Why does this matter?
- ChatGPT has still managed to gain a remarkable level of popularity, despite low usage. This suggests that even though not many people are using it, they are aware of it and its potential capabilities. More people reported using ChatGPT for entertainment or to educate themselves rather than for work.\n- People anticipate AI to have a greater impact on jobs such as software engineers, graphic designers, and journalists. But the expectation is that AI as a whole, not just ChatGPT, will be the driving force behind this.\n- Concern about AI is increasing, not decreasing. 47% of respondents said AI makes them more worried than excited, comparing to 31% last year. This concern seems to rise with the level of AI knowledge one possesses.
Industries unshaken by AI:
- As per the survey, employed individuals who are aware of ChatGPT don't see it drastically affecting their jobs. The sectors like hospitality, entertainment, construction, and manufacturing feel the least threatened.
Stay updated about AI and its influence on different verticals!
Don't miss out on the latest insights, developments, and trajectories of AI. Our free newsletter is all you need to be au fait with the AI world: supercharged-ai.beehiiv.com/subscribe
(source: techcrunch.com/2023/08/28/survey-finds-relatively-few-americans-actually-use-or-fear-chatgpt)"},"title":{"matchLevel":"none","matchedWords":[],"value":"ChatGPT usage remains low, suggests Pew Research"}},"_tags":["story","author_bloxygen","story_37303714","ask_hn"],"author":"bloxygen","children":[37303817],"created_at":"2023-08-29T05:02:08Z","created_at_i":1693285328,"num_comments":1,"objectID":"37303714","points":1,"story_id":37303714,"story_text":"The usage and fear surrounding ChatGPT, a language model developed by AI, aren't as prevalent as you might think, according to a recent poll from Pew Research. Only 18% of Americans have reportedly used ChatGPT. The demographic that uses it the most? Men aged 18-29 that are educated in college, but even that's just a 30-40% usage rate.
Why does this matter?
- ChatGPT has still managed to gain a remarkable level of popularity, despite low usage. This suggests that even though not many people are using it, they are aware of it and its potential capabilities. More people reported using ChatGPT for entertainment or to educate themselves rather than for work.\n- People anticipate AI to have a greater impact on jobs such as software engineers, graphic designers, and journalists. But the expectation is that AI as a whole, not just ChatGPT, will be the driving force behind this.\n- Concern about AI is increasing, not decreasing. 47% of respondents said AI makes them more worried than excited, comparing to 31% last year. This concern seems to rise with the level of AI knowledge one possesses.
Industries unshaken by AI:
- As per the survey, employed individuals who are aware of ChatGPT don't see it drastically affecting their jobs. The sectors like hospitality, entertainment, construction, and manufacturing feel the least threatened.
Stay updated about AI and its influence on different verticals!
Don't miss out on the latest insights, developments, and trajectories of AI. Our free newsletter is all you need to be au fait with the AI world: supercharged-ai.beehiiv.com/subscribe
(source: techcrunch.com/2023/08/28/survey-finds-relatively-few-americans-actually-use-or-fear-chatgpt)","title":"ChatGPT usage remains low, suggests Pew Research","updated_at":"2024-09-20T14:56:27Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"hhs"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Popular AI chatbots found to give error-ridden legal answers"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"https://news.bloomberglaw.com/business-and-practice/legal-errors-by-top-ai-models-alarmingly-prevalent-study-says"}},"_tags":["story","author_hhs","story_38969581"],"author":"hhs","created_at":"2024-01-12T15:56:12Z","created_at_i":1705074972,"num_comments":0,"objectID":"38969581","points":2,"story_id":38969581,"title":"Popular AI chatbots found to give error-ridden legal answers","updated_at":"2024-09-20T16:07:11Z","url":"https://news.bloomberglaw.com/business-and-practice/legal-errors-by-top-ai-models-alarmingly-prevalent-study-says"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"flyx"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Hey HN community -
I\u2019m Ivan from Datasaur (https://datasaur.ai/) - we build software to allow humans to more efficiently label data for training natural language processing (NLP).
NLP algorithms are being trained in a wide variety of industries - from customer service to legal contracts, forum moderation to restaurant reviews. All these algorithms benefit from recent breakthroughs in academia and a generous open-source community. However, in order to be deployed to the real world, they require a custom set of training data to learn and understand the language unique to each industry. Therefore, people around the world are meticulously labeling data samples.
Example sentence: London is the capital and largest city of England and of the United Kingdom.
Labels: \u201cLondon\u201d \u2014> \u201ccapital\u201d, \u201cUnited Kingdom\u201d
Labels: \u201cLondon\u201d \u2014> \u201clargest city\u201d, \u201cEngland\u201d
In the last few years I\u2019ve worked at companies such as Apple and Yahoo and noticed that many organizations tend to reinvent the wheel when creating labeling interfaces for their labelers. Some companies still do this work in Excel.\nWe saw an opportunity to create a "single interface to rule them all" - to handle all sorts of text labeling tasks.
We leverage existing NLP capabilities to intelligently validate the quality of labels in a document and complement human judgment. Furthermore, we already understand terms like \u201cStarbucks\u201d and \u201cNew York\u201d - why spend time labeling these terms from scratch every time? We created an API so you can plug in existing models to apply a first pass on labeling the document. We also built many other extensions to help labelers optimize their time - a \u201cfind and label\u201d extension for labeling repetitive terms, a dictionary extension for quickly looking up unfamiliar terms. We spent the past year building out the labeling solution I wish I could have used.
We now handle named entity recognition, parts of speech, document labeling, coreference resolution (multiple words referring to the same object/person) and dependency parsing (drawing relationships between words). A case study with one of our clients shows 70% improved labeling efficiency upon adopting the Datasaur platform, and we have much more room to improve.
We also spoken with 100+ AI teams globally and identified the best practices in labeling. In addition to providing an enhanced interface, we can help track labeler performance, peer disagreement scores, and detect/remove labeler bias. By incorporating and encoding these features into our software, we can not only help improve the labeling efficiency but also improve the quality of the data and therefore the resulting AI model.
We believe that as AI becomes ever more prevalent and ubiquitous, labeling will become an increasingly important task. AI is a garbage-in, garbage-out technology, and the quantity and quality of data can often make a critical difference in the resulting AI model. We\u2019re really excited to open Datasaur up to the world today and hear your feedback. Have you run into similar labeling issues? What tips and tricks have you employed to keep up with AI\u2019s voracious appetite for data? We\u2019d love to hear how you\u2019ve tackled data labeling at your own companies. Thanks so much in advance!
Ivan"},"title":{"matchLevel":"none","matchedWords":[],"value":"Launch HN: Datasaur (YC W20) \u2013 data labeling interface for NLP"}},"_tags":["story","author_flyx","story_22506429","launch_hn"],"author":"flyx","children":[22506563,22506750,22506772,22506802,22506896,22506909,22506925,22506970,22507009,22507037,22507104,22507111,22507250,22507263,22507331,22507337,22507410,22507423,22507456,22507460,22507621,22507779,22507981,22508316,22508958,22508981,22509090,22509275,22509415,22510247,22510535,22524861,22601811],"created_at":"2020-03-06T19:25:49Z","created_at_i":1583522749,"num_comments":62,"objectID":"22506429","points":174,"story_id":22506429,"story_text":"Hey HN community -
I\u2019m Ivan from Datasaur (https://datasaur.ai/) - we build software to allow humans to more efficiently label data for training natural language processing (NLP).
NLP algorithms are being trained in a wide variety of industries - from customer service to legal contracts, forum moderation to restaurant reviews. All these algorithms benefit from recent breakthroughs in academia and a generous open-source community. However, in order to be deployed to the real world, they require a custom set of training data to learn and understand the language unique to each industry. Therefore, people around the world are meticulously labeling data samples.
Example sentence: London is the capital and largest city of England and of the United Kingdom.
Labels: \u201cLondon\u201d \u2014> \u201ccapital\u201d, \u201cUnited Kingdom\u201d
Labels: \u201cLondon\u201d \u2014> \u201clargest city\u201d, \u201cEngland\u201d
In the last few years I\u2019ve worked at companies such as Apple and Yahoo and noticed that many organizations tend to reinvent the wheel when creating labeling interfaces for their labelers. Some companies still do this work in Excel.\nWe saw an opportunity to create a "single interface to rule them all" - to handle all sorts of text labeling tasks.
We leverage existing NLP capabilities to intelligently validate the quality of labels in a document and complement human judgment. Furthermore, we already understand terms like \u201cStarbucks\u201d and \u201cNew York\u201d - why spend time labeling these terms from scratch every time? We created an API so you can plug in existing models to apply a first pass on labeling the document. We also built many other extensions to help labelers optimize their time - a \u201cfind and label\u201d extension for labeling repetitive terms, a dictionary extension for quickly looking up unfamiliar terms. We spent the past year building out the labeling solution I wish I could have used.
We now handle named entity recognition, parts of speech, document labeling, coreference resolution (multiple words referring to the same object/person) and dependency parsing (drawing relationships between words). A case study with one of our clients shows 70% improved labeling efficiency upon adopting the Datasaur platform, and we have much more room to improve.
We also spoken with 100+ AI teams globally and identified the best practices in labeling. In addition to providing an enhanced interface, we can help track labeler performance, peer disagreement scores, and detect/remove labeler bias. By incorporating and encoding these features into our software, we can not only help improve the labeling efficiency but also improve the quality of the data and therefore the resulting AI model.
We believe that as AI becomes ever more prevalent and ubiquitous, labeling will become an increasingly important task. AI is a garbage-in, garbage-out technology, and the quantity and quality of data can often make a critical difference in the resulting AI model. We\u2019re really excited to open Datasaur up to the world today and hear your feedback. Have you run into similar labeling issues? What tips and tricks have you employed to keep up with AI\u2019s voracious appetite for data? We\u2019d love to hear how you\u2019ve tackled data labeling at your own companies. Thanks so much in advance!
Ivan","title":"Launch HN: Datasaur (YC W20) \u2013 data labeling interface for NLP","updated_at":"2024-09-20T05:47:03Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"IAmNeo"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"I just had this idea, you read it all the time AI slop is so prevalent people are getting banned for a year for submitting science papers to arXiv with it, moans of angst from developers, even Microsoft doing its own study where AI degrades the quality of simple documents, and the beloved em-dash.
I don't really have the know-how or the time but it occurred to me, if we created a public data set that could be submitted to publicly, we could catalog and organize all the AI slop, the different types, with explanations about why it is slop and why not to do it, and then train a large language model using this data set included, to help correct itself.
I don't really know the technical details of training a large language model,is this even possible?"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Slop Bucket Idea \u2013 a dataset of AI slop (train AI what not to do)"}},"_tags":["story","author_IAmNeo","story_48174947","ask_hn"],"author":"IAmNeo","children":[48174966,48174987],"created_at":"2026-05-18T02:11:19Z","created_at_i":1779070279,"num_comments":4,"objectID":"48174947","points":2,"story_id":48174947,"story_text":"I just had this idea, you read it all the time AI slop is so prevalent people are getting banned for a year for submitting science papers to arXiv with it, moans of angst from developers, even Microsoft doing its own study where AI degrades the quality of simple documents, and the beloved em-dash.
I don't really have the know-how or the time but it occurred to me, if we created a public data set that could be submitted to publicly, we could catalog and organize all the AI slop, the different types, with explanations about why it is slop and why not to do it, and then train a large language model using this data set included, to help correct itself.
I don't really know the technical details of training a large language model,is this even possible?","title":"Slop Bucket Idea \u2013 a dataset of AI slop (train AI what not to do)","updated_at":"2026-05-18T03:42:23Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"titusblair"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Hey HN,
We\u2019ve been working in-house on a platform that tests the security of chatbots and voicebots by intentionally trying to break them.
As AI-driven bots become more prevalent across sectors like customer service, healthcare, and finance, ensuring they are secure from exploitation is critical. Many companies focus on training their AI to perform well but often overlook the necessity of breaking them to identify vulnerabilities\u2014essential to ensuring their robustness in the real world.
Why We Built This:
We realized how easily AI models could be manipulated through adversarial inputs and social engineering tactics. With the rise of chatbots and voicebots in sensitive areas, traditional testing methods fell short.
What We Did:
We developed an in-house platform (code named RedOps) that simulates real-world attacks on chatbots and voicebots, including:
1. Contextual Manipulation: Testing how the bot handles changes in conversation context or ambiguous input.\n2. Adversarial Attacks: Feeding in slightly altered inputs designed to trick the bot into revealing sensitive information.\n3. Ethical Compliance: Ensuring that the bot doesn\u2019t produce biased, harmful, or inappropriate content.\n4. Polymorphic Testing: Submitting the same question in various forms to see if the bot responds consistently and securely.\n5. Social Engineering: Simulating how an attacker might try to extract sensitive information by posing as a trusted user.
Key Findings:
1. Context is Everything:\nExample: We started a conversation with a chatbot about the weather, then subtly shifted to privacy. The bot, trying to be helpful, ended up revealing previous user inputs because it failed to recognize the context change.
Lesson: Bots must be trained to recognize shifts into sensitive contexts and should refuse to divulge sensitive information without proper validation.
Fix: Implement context-detection mechanisms, context reset protocols, and update prompts to include fallbacks or refusals for sensitive topics.
2. Biases Lurk in Unexpected Places:\nExample: In a test, a voicebot displayed bias when asked about public figures, based on data it had been trained on. This bias emerged only when specific questions were asked in sequence.
Lesson: Regular audits and retraining are essential to minimize biases. Prompt engineering plays a crucial role in guiding bots toward neutral and ethical responses.
Fix: Use automated bias detection tools, retrain models with diversified datasets, and calibrate prompts to be more neutral, including disclaimers for subjective topics.
3. Security is a Moving Target:\nExample: A chatbot that previously passed security audits became vulnerable after an update introduced a new feature. This feature enhanced user interaction but inadvertently opened a new vulnerability.
Lesson: Continuous security testing is crucial as AI evolves. Regularly update security protocols and test against the latest threats.
Fix: Implement automated regression tests, set up continuous monitoring, and update prompts to include safety checks for risky actions.
Free Security Test + Detailed Analysis:
As a way of giving back to the community, we\u2019re offering a free security test of your chatbot or voicebot. If you have a bot in production, send a link to redops@primemindai.com. We\u2019ll run it through our platform and provide you with a detailed report of our findings.
Here\u2019s What You\u2019ll Get:
1. Vulnerability Report: Detailed security issues identified.\n2. Impact Analysis: Potential risks associated with each vulnerability.\n3. Actionable Tips: Specific recommendations to improve security and prevent future attacks.\n4. Prevention Strategies: Guidance on fortifying your bot against real-world attacks.
We\u2019d love to hear your thoughts. Have you faced similar challenges with AI security? How do you approach securing chatbots and voicebots?"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"We Built a Tool to Hack Our Own AI: Lessons Learned Securing Chatbots/Voicebots"}},"_tags":["story","author_titusblair","story_41249044","ask_hn"],"author":"titusblair","children":[41251457],"created_at":"2024-08-14T18:25:59Z","created_at_i":1723659959,"num_comments":2,"objectID":"41249044","points":16,"story_id":41249044,"story_text":"Hey HN,
We\u2019ve been working in-house on a platform that tests the security of chatbots and voicebots by intentionally trying to break them.
As AI-driven bots become more prevalent across sectors like customer service, healthcare, and finance, ensuring they are secure from exploitation is critical. Many companies focus on training their AI to perform well but often overlook the necessity of breaking them to identify vulnerabilities\u2014essential to ensuring their robustness in the real world.
Why We Built This:
We realized how easily AI models could be manipulated through adversarial inputs and social engineering tactics. With the rise of chatbots and voicebots in sensitive areas, traditional testing methods fell short.
What We Did:
We developed an in-house platform (code named RedOps) that simulates real-world attacks on chatbots and voicebots, including:
1. Contextual Manipulation: Testing how the bot handles changes in conversation context or ambiguous input.\n2. Adversarial Attacks: Feeding in slightly altered inputs designed to trick the bot into revealing sensitive information.\n3. Ethical Compliance: Ensuring that the bot doesn\u2019t produce biased, harmful, or inappropriate content.\n4. Polymorphic Testing: Submitting the same question in various forms to see if the bot responds consistently and securely.\n5. Social Engineering: Simulating how an attacker might try to extract sensitive information by posing as a trusted user.
Key Findings:
1. Context is Everything:\nExample: We started a conversation with a chatbot about the weather, then subtly shifted to privacy. The bot, trying to be helpful, ended up revealing previous user inputs because it failed to recognize the context change.
Lesson: Bots must be trained to recognize shifts into sensitive contexts and should refuse to divulge sensitive information without proper validation.
Fix: Implement context-detection mechanisms, context reset protocols, and update prompts to include fallbacks or refusals for sensitive topics.
2. Biases Lurk in Unexpected Places:\nExample: In a test, a voicebot displayed bias when asked about public figures, based on data it had been trained on. This bias emerged only when specific questions were asked in sequence.
Lesson: Regular audits and retraining are essential to minimize biases. Prompt engineering plays a crucial role in guiding bots toward neutral and ethical responses.
Fix: Use automated bias detection tools, retrain models with diversified datasets, and calibrate prompts to be more neutral, including disclaimers for subjective topics.
3. Security is a Moving Target:\nExample: A chatbot that previously passed security audits became vulnerable after an update introduced a new feature. This feature enhanced user interaction but inadvertently opened a new vulnerability.
Lesson: Continuous security testing is crucial as AI evolves. Regularly update security protocols and test against the latest threats.
Fix: Implement automated regression tests, set up continuous monitoring, and update prompts to include safety checks for risky actions.
Free Security Test + Detailed Analysis:
As a way of giving back to the community, we\u2019re offering a free security test of your chatbot or voicebot. If you have a bot in production, send a link to redops@primemindai.com. We\u2019ll run it through our platform and provide you with a detailed report of our findings.
Here\u2019s What You\u2019ll Get:
1. Vulnerability Report: Detailed security issues identified.\n2. Impact Analysis: Potential risks associated with each vulnerability.\n3. Actionable Tips: Specific recommendations to improve security and prevent future attacks.\n4. Prevention Strategies: Guidance on fortifying your bot against real-world attacks.
We\u2019d love to hear your thoughts. Have you faced similar challenges with AI security? How do you approach securing chatbots and voicebots?","title":"We Built a Tool to Hack Our Own AI: Lessons Learned Securing Chatbots/Voicebots","updated_at":"2024-09-20T17:38:26Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"olokobayusuf"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Hey all,
You've probably seen projects that add objects to an image from a style or text prompt, like InteriorAI (levelsio) and Adobe Firefly. The prevalent issue with these diffusion-based inpainting approaches is that they don't yet have great conditioning on lighting, perspective, and structure. You'll often get incorrect or generic shadows; warped-looking objects; and distorted backgrounds.
What is Fill 3D?\nFill 3D is an exploration on doing generative fill in 3D to render ultra-realistic results that harmonize with the background image, using industry-standard path tracing, akin to compositing in Hollywood movies.
How does it work?\n1. Deproject: First, deproject an image to a 3D shell using both geometric and photometric cues from the input image.\n2. Place: Draw rectangles and describe what you want in them, akin to Photoshop's Generative Fill feature.\n3. Render: Use good ol' path tracing to render ultra-realistic results.
Why Fill 3D?\n+ The results are insanely realistic (see video in the github repo, or on the website).\n+ Fast enough: Currently, generations take 40-80 seconds. Diffusion takes ~10seconds, so we're slower, but for the level of realism, it's pretty good.\n+ Potential applications: I'm thinking of virtual staging in real estate media, what do you think?
Check it out at https://fill3d.ai\n+ There's API access! :D\n+ Right now, you need an image of an empty room. Will loosen this restriction over time.
Fill 3D is built on Function (https://fxn.ai). With Function, I can run the Python functions that do the steps above on powerful GPUs with only code (no Dockerfile, YAML, k8s, etc), and invoke them from just about anywhere. I'm the founder of fxn.
Tell me what you think!!
PS: This is my first Show HN, so please be nice :)"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Show HN: Generative Fill with AI and 3D"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/fill3d/fill"}},"_tags":["story","author_olokobayusuf","story_37695530","show_hn"],"author":"olokobayusuf","children":[37696047,37696085,37696086,37696090,37696141,37696349,37696386,37696405,37696538,37696591,37696748,37696887,37696933,37696936,37696995,37696997,37697095,37697132,37697292,37697858,37698041,37698212,37698310,37698339,37698824,37698953,37700057,37700810,37700814,37701993,37702592,37702805,37703570,37719128],"created_at":"2023-09-28T20:41:30Z","created_at_i":1695933690,"num_comments":102,"objectID":"37695530","points":360,"story_id":37695530,"story_text":"Hey all,
You've probably seen projects that add objects to an image from a style or text prompt, like InteriorAI (levelsio) and Adobe Firefly. The prevalent issue with these diffusion-based inpainting approaches is that they don't yet have great conditioning on lighting, perspective, and structure. You'll often get incorrect or generic shadows; warped-looking objects; and distorted backgrounds.
What is Fill 3D?\nFill 3D is an exploration on doing generative fill in 3D to render ultra-realistic results that harmonize with the background image, using industry-standard path tracing, akin to compositing in Hollywood movies.
How does it work?\n1. Deproject: First, deproject an image to a 3D shell using both geometric and photometric cues from the input image.\n2. Place: Draw rectangles and describe what you want in them, akin to Photoshop's Generative Fill feature.\n3. Render: Use good ol' path tracing to render ultra-realistic results.
Why Fill 3D?\n+ The results are insanely realistic (see video in the github repo, or on the website).\n+ Fast enough: Currently, generations take 40-80 seconds. Diffusion takes ~10seconds, so we're slower, but for the level of realism, it's pretty good.\n+ Potential applications: I'm thinking of virtual staging in real estate media, what do you think?
Check it out at https://fill3d.ai\n+ There's API access! :D\n+ Right now, you need an image of an empty room. Will loosen this restriction over time.
Fill 3D is built on Function (https://fxn.ai). With Function, I can run the Python functions that do the steps above on powerful GPUs with only code (no Dockerfile, YAML, k8s, etc), and invoke them from just about anywhere. I'm the founder of fxn.
Tell me what you think!!
PS: This is my first Show HN, so please be nice :)","title":"Show HN: Generative Fill with AI and 3D","updated_at":"2025-08-15T21:48:21Z","url":"https://github.com/fill3d/fill"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"supremus_58"},"story_text":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["prevalent"],"value":"I dropped out of school a few years ago and have been working as a developer. I plan to start attending school again in the coming fall. \nMy question: what is the best route of study to become a master of all things machine learning? \nMy plan is to go in as a double major for math and computer science as I was before. Which topics in mathematics are most prevalent for machine learning and more specifically neural networks?"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Ask HN: What to study to become a machine learning/ai master?"}},"_tags":["story","author_supremus_58","story_13611095","ask_hn"],"author":"supremus_58","children":[13611588,13611666,13611803,13613551,13615539,13633818],"created_at":"2017-02-09T21:21:01Z","created_at_i":1486675261,"num_comments":10,"objectID":"13611095","points":24,"story_id":13611095,"story_text":"I dropped out of school a few years ago and have been working as a developer. I plan to start attending school again in the coming fall. \nMy question: what is the best route of study to become a master of all things machine learning? \nMy plan is to go in as a double major for math and computer science as I was before. Which topics in mathematics are most prevalent for machine learning and more specifically neural networks?","title":"Ask HN: What to study to become a machine learning/ai master?","updated_at":"2024-09-20T00:21:06Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"duckerduck"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Hi HN, like many I've been interested in the direction software engineering is taking now that coding LLMs are becoming prevalent. It seems that we're not quite there for "natural language programming", but it seems new abstractions are already starting to form. In order to explore this further I've built semcheck (semantic checker). It's a simple cli tool that can be used in CI or pre-commit to check that your implementation matches your specification using LLMs.
The inspiration came while I was working on another project where I needed a data structure for a GeoJSON object, I passed Claude the text of RFC-7946 and it gave me an implementation. It took some back and forth after that before I was happy with it, but this also meant the RFC went out of context for the LLM. That's why I asked Claude again to check the RFC to make sure we haven't strayed too far from the spec. It occurred to me that it would be good to have a formal way of defining these kinds of checks that can be run in a pre-commit or merge request flow.
Creating this tool was itself an experiment to try "spec-driven-development" using Claude Code, a middle ground between completely vibe-coding and traditional programming. My workflow was as follows: ask AI to write a spec and implementation plan, edit these manually to my liking, then ask AI to execute one step at a time. Being careful that the AI doesn't drift too far from what I think is required. My very first commit [1] is the specification of the config file structure and an implementation plan.
As soon as semcheck was in a state where it could check itself it started to find issues [2]. I found that this workflow improves not just your implementation but helps you refine your specification at the same time.
Besides specification, I also started to include documentation in my rules, making sure that the configuration examples and CLI flags I have in my README.md file stay in line with implementation [3].
The best thing is that you can put found issues directly back into your AI editor for a quick iteration cycle.
Some learnings:
- LLMs are very good at finding discrepancies, as long as the number of files you pass to the comparison function isn't too large, in other words the true-positive results are quite good.
- False-positives: the LLM is a know-it-all (literally) and often thinks it knows better. The LLM is eager to use its own world knowledge to find faults. This can both be nice and problematic. I've often had it complain that my Go version doesn't exist, but it was simply released after the knowledge cutoff of that model. I specifically prompt [4] the model to only find discrepancies, but it often "chooses" to use its knowledge anyway.
- In an effort to reduce false-positives I ask the model to give me a confidence score (0-1), to indicate to me how sure it was that the issue it found is actually applicable in this scenario. The models are always super confident and output values > 0.7 almost exclusively.
- One thing that did reduced false-positives significantly is asking the model to give its reasoning before assigning a severity level to an issue found.
- In my (rudimentary) experiments I found that "thinking" models like O3 don't improve on performance much and are not worth the additional tokens/time. (likely because I already ask for the reasoning anyway)
- The models that perform best are Claude 4 and GPT-4.1
Let me know if you could see this be useful in your workflow, and what feature you would need to make it functional.
[1]: https://github.com/rejot-dev/semcheck/commit/ce0af27ca0077fe...
[2]: https://github.com/rejot-dev/semcheck/commit/2f96fc428b551d9...
[3]: https://github.com/rejot-dev/semcheck/blob/47f7aaf98811c54e2...
[4]: https://github.com/rejot-dev/semcheck/blob/fec2df48304d9eff9..."},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Show HN: Semcheck \u2013 AI Tool for checking implementation follows spec"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/rejot-dev/semcheck"}},"_tags":["story","author_duckerduck","story_44432215","show_hn"],"author":"duckerduck","created_at":"2025-07-01T09:49:26Z","created_at_i":1751363366,"num_comments":0,"objectID":"44432215","points":19,"story_id":44432215,"story_text":"Hi HN, like many I've been interested in the direction software engineering is taking now that coding LLMs are becoming prevalent. It seems that we're not quite there for "natural language programming", but it seems new abstractions are already starting to form. In order to explore this further I've built semcheck (semantic checker). It's a simple cli tool that can be used in CI or pre-commit to check that your implementation matches your specification using LLMs.
The inspiration came while I was working on another project where I needed a data structure for a GeoJSON object, I passed Claude the text of RFC-7946 and it gave me an implementation. It took some back and forth after that before I was happy with it, but this also meant the RFC went out of context for the LLM. That's why I asked Claude again to check the RFC to make sure we haven't strayed too far from the spec. It occurred to me that it would be good to have a formal way of defining these kinds of checks that can be run in a pre-commit or merge request flow.
Creating this tool was itself an experiment to try "spec-driven-development" using Claude Code, a middle ground between completely vibe-coding and traditional programming. My workflow was as follows: ask AI to write a spec and implementation plan, edit these manually to my liking, then ask AI to execute one step at a time. Being careful that the AI doesn't drift too far from what I think is required. My very first commit [1] is the specification of the config file structure and an implementation plan.
As soon as semcheck was in a state where it could check itself it started to find issues [2]. I found that this workflow improves not just your implementation but helps you refine your specification at the same time.
Besides specification, I also started to include documentation in my rules, making sure that the configuration examples and CLI flags I have in my README.md file stay in line with implementation [3].
The best thing is that you can put found issues directly back into your AI editor for a quick iteration cycle.
Some learnings:
- LLMs are very good at finding discrepancies, as long as the number of files you pass to the comparison function isn't too large, in other words the true-positive results are quite good.
- False-positives: the LLM is a know-it-all (literally) and often thinks it knows better. The LLM is eager to use its own world knowledge to find faults. This can both be nice and problematic. I've often had it complain that my Go version doesn't exist, but it was simply released after the knowledge cutoff of that model. I specifically prompt [4] the model to only find discrepancies, but it often "chooses" to use its knowledge anyway.
- In an effort to reduce false-positives I ask the model to give me a confidence score (0-1), to indicate to me how sure it was that the issue it found is actually applicable in this scenario. The models are always super confident and output values > 0.7 almost exclusively.
- One thing that did reduced false-positives significantly is asking the model to give its reasoning before assigning a severity level to an issue found.
- In my (rudimentary) experiments I found that "thinking" models like O3 don't improve on performance much and are not worth the additional tokens/time. (likely because I already ask for the reasoning anyway)
- The models that perform best are Claude 4 and GPT-4.1
Let me know if you could see this be useful in your workflow, and what feature you would need to make it functional.
[1]: https://github.com/rejot-dev/semcheck/commit/ce0af27ca0077fe...
[2]: https://github.com/rejot-dev/semcheck/commit/2f96fc428b551d9...
[3]: https://github.com/rejot-dev/semcheck/blob/47f7aaf98811c54e2...
[4]: https://github.com/rejot-dev/semcheck/blob/fec2df48304d9eff9...","title":"Show HN: Semcheck \u2013 AI Tool for checking implementation follows spec","updated_at":"2025-11-13T14:56:02Z","url":"https://github.com/rejot-dev/semcheck"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"cchio"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["prevalent"],"value":"Why adversarial examples are more prevalent in higher dimensional data"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://aivillage.org/posts/dimensionality-and-adversarial/"}},"_tags":["story","author_cchio","story_17130523"],"author":"cchio","created_at":"2018-05-23T00:08:55Z","created_at_i":1527034135,"num_comments":0,"objectID":"17130523","points":9,"story_id":17130523,"title":"Why adversarial examples are more prevalent in higher dimensional data","updated_at":"2024-09-20T02:29:42Z","url":"https://aivillage.org/posts/dimensionality-and-adversarial/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"OnlineInference"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["prevalent"],"value":"ReLU vs. Sigmoid Function in Deep Neural Networks: Why ReLU Is So Prevalent"},"url":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"https://wandb.ai/ayush-thakur/dl-question-bank/reports/ReLU-vs-Sigmoid-Function-in-Deep-Neural-Networks-Why-ReLU-is-so-Prevalent--VmlldzoyMDk0MzI"}},"_tags":["story","author_OnlineInference","story_27764017"],"author":"OnlineInference","created_at":"2021-07-07T18:03:20Z","created_at_i":1625681000,"num_comments":0,"objectID":"27764017","points":4,"story_id":27764017,"title":"ReLU vs. Sigmoid Function in Deep Neural Networks: Why ReLU Is So Prevalent","updated_at":"2024-09-20T08:56:19Z","url":"https://wandb.ai/ayush-thakur/dl-question-bank/reports/ReLU-vs-Sigmoid-Function-in-Deep-Neural-Networks-Why-ReLU-is-so-Prevalent--VmlldzoyMDk0MzI"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"zachrsweedler"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Introducing Buybase (https://www.buybase.ai)
Natural language search and purchasing have the potential to render traditional online storefronts obsolete, as conversational AI agents with text and voice modalities become more prevalent. This means, having users manually search, apply filters on search result pages, add items to cart, and fill out checkout forms become a medium of the past. This shift ushers in a new era of "natural-language purchasing." With some predications saying agents armed with digital wallets will represent the majority of online transactions by 2030.
Buybase is positioned as the infrastructure for purchasing agents or as some may call "AI shopping assistants."
We offer a developer API designed to search, personalize, recommend, and purchase products from various marketplaces (Amazon, Walmart, Target, Shopify, etc) on behalf of your users. Our solution is not bound by official third-party APIs, is built to be LLM-first, and offers a way for developers to capture a percent of each transaction for monetizing their apps.
Buybase\u2019s API offers several features to support developers purchasing agent applications:
\u2022 Authentication: Manages user account session data, bot detection, two-factor authentication, and captchas.
\u2022 Monetization: Provides infrastructure to collect transaction fees from users, allowing developers to earn a percentage of each transaction.
\u2022 Maintenance: Maintains search and purchasing automation with e-commerce platforms.
\u2022 Personalization: Tracks user preferences and order history to enhance search results.
\u2022 Developer-Friendly: Offers comprehensive documentation for quick integration.
\u2022 Security: Implements a zero-trust architecture and encrypted user session storage.
Right now, we are in private beta, offer Amazon search and purchasing on behalf of users, and are actively looking for builders and developers to join and provide feedback. If you're interested in joining, please go to https://buybase.ai and click "Join API Beta"!
Documentation: https://docs.buybase.ai | \nDiscord Community: https://lnkd.in/erxST48R"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Show HN: Buybase \u2013 Ecom for your AI agent"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"https://www.buybase.ai/"}},"_tags":["story","author_zachrsweedler","story_42579471","show_hn"],"author":"zachrsweedler","created_at":"2025-01-02T22:08:36Z","created_at_i":1735855716,"num_comments":0,"objectID":"42579471","points":3,"story_id":42579471,"story_text":"Introducing Buybase (https://www.buybase.ai)
Natural language search and purchasing have the potential to render traditional online storefronts obsolete, as conversational AI agents with text and voice modalities become more prevalent. This means, having users manually search, apply filters on search result pages, add items to cart, and fill out checkout forms become a medium of the past. This shift ushers in a new era of "natural-language purchasing." With some predications saying agents armed with digital wallets will represent the majority of online transactions by 2030.
Buybase is positioned as the infrastructure for purchasing agents or as some may call "AI shopping assistants."
We offer a developer API designed to search, personalize, recommend, and purchase products from various marketplaces (Amazon, Walmart, Target, Shopify, etc) on behalf of your users. Our solution is not bound by official third-party APIs, is built to be LLM-first, and offers a way for developers to capture a percent of each transaction for monetizing their apps.
Buybase\u2019s API offers several features to support developers purchasing agent applications:
\u2022 Authentication: Manages user account session data, bot detection, two-factor authentication, and captchas.
\u2022 Monetization: Provides infrastructure to collect transaction fees from users, allowing developers to earn a percentage of each transaction.
\u2022 Maintenance: Maintains search and purchasing automation with e-commerce platforms.
\u2022 Personalization: Tracks user preferences and order history to enhance search results.
\u2022 Developer-Friendly: Offers comprehensive documentation for quick integration.
\u2022 Security: Implements a zero-trust architecture and encrypted user session storage.
Right now, we are in private beta, offer Amazon search and purchasing on behalf of users, and are actively looking for builders and developers to join and provide feedback. If you're interested in joining, please go to https://buybase.ai and click "Join API Beta"!
Documentation: https://docs.buybase.ai | \nDiscord Community: https://lnkd.in/erxST48R","title":"Show HN: Buybase \u2013 Ecom for your AI agent","updated_at":"2025-01-14T22:44:40Z","url":"https://www.buybase.ai/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"jinmingjian"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Since the global launch of JoinBase in October 2022, we have received overwhelming requests from totally 33 users applied our free distribution on the website. We are exciting that we can help the world.
Unlike the currently prevalent, over-sized cloud platforms, we aspire to construct an AIoT infrastructure that is "everyone-accessible", from the ground up. It doesn't require specialized knowledg; it utilizes cutting-edge hardware, contemporary kernels, and state-of-the-art database and compilation technologies; it works on several-dollars-cheap single-board computers, and also scaled to the mega-threads and clusters. Despite there is still much room for growth, we are proud of what we have accomplished so far.
There's a release blog to the free full-functional community version to try, if you're interested.
:thanks:"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Show HN: JoinBase, single binary AIoT-first data-service platform"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://joinbase.io/blog/joinbase-2023/"}},"_tags":["story","author_jinmingjian","story_34719323","show_hn"],"author":"jinmingjian","children":[34719384],"created_at":"2023-02-09T03:14:37Z","created_at_i":1675912477,"num_comments":0,"objectID":"34719323","points":3,"story_id":34719323,"story_text":"Since the global launch of JoinBase in October 2022, we have received overwhelming requests from totally 33 users applied our free distribution on the website. We are exciting that we can help the world.
Unlike the currently prevalent, over-sized cloud platforms, we aspire to construct an AIoT infrastructure that is "everyone-accessible", from the ground up. It doesn't require specialized knowledg; it utilizes cutting-edge hardware, contemporary kernels, and state-of-the-art database and compilation technologies; it works on several-dollars-cheap single-board computers, and also scaled to the mega-threads and clusters. Despite there is still much room for growth, we are proud of what we have accomplished so far.
There's a release blog to the free full-functional community version to try, if you're interested.
:thanks:","title":"Show HN: JoinBase, single binary AIoT-first data-service platform","updated_at":"2024-09-20T13:12:00Z","url":"https://joinbase.io/blog/joinbase-2023/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"elocinstr8t"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"With automation becoming more and more prevalent these days, some people fear about automation and AI replacing us. I guess in this sense developers are safe but does it threaten you? Why or why not?"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Do AI/automation threaten you?"}},"_tags":["story","author_elocinstr8t","story_18456787","ask_hn"],"author":"elocinstr8t","children":[18456792,18457436],"created_at":"2018-11-15T04:07:30Z","created_at_i":1542254850,"num_comments":4,"objectID":"18456787","points":2,"story_id":18456787,"story_text":"With automation becoming more and more prevalent these days, some people fear about automation and AI replacing us. I guess in this sense developers are safe but does it threaten you? Why or why not?","title":"Do AI/automation threaten you?","updated_at":"2024-09-20T03:20:32Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"alexliu518"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"As we delve deeper into the 21st century, the intersection of artificial intelligence (AI) and human creativity is becoming an increasingly prevalent and provocative topic. With the advent of advanced AI tools capable of generating art, music, literature, and even code, we stand at the precipice of a new era where the boundaries between human and machine-generated creativity are becoming blurred.
This fusion raises several compelling questions:
Can AI truly be creative, or is it merely mimicking patterns it has learned from vast datasets?\nWhat does the collaboration between AI and human artists mean for the future of creative professions?\nHow do we navigate the ethical and intellectual property concerns that arise with AI-generated content?\nMoreover, as AI tools become more accessible and integrated into creative workflows, how will our perception of art and originality evolve? Will the democratization of creative tools lead to a renaissance of human expression, or will it dilute the value of individual creativity?
I invite you all to share your insights, concerns, and predictions on this fascinating intersection of technology and human ingenuity. How do you envision the role of AI in the creative process, and what implications does this hold for the future of art, culture, and technology?"},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"The Fusion of AI and Human Creativity: Navigating the New Frontier"}},"_tags":["story","author_alexliu518","story_39812688","ask_hn"],"author":"alexliu518","created_at":"2024-03-25T03:42:35Z","created_at_i":1711338155,"num_comments":0,"objectID":"39812688","points":2,"story_id":39812688,"story_text":"As we delve deeper into the 21st century, the intersection of artificial intelligence (AI) and human creativity is becoming an increasingly prevalent and provocative topic. With the advent of advanced AI tools capable of generating art, music, literature, and even code, we stand at the precipice of a new era where the boundaries between human and machine-generated creativity are becoming blurred.
This fusion raises several compelling questions:
Can AI truly be creative, or is it merely mimicking patterns it has learned from vast datasets?\nWhat does the collaboration between AI and human artists mean for the future of creative professions?\nHow do we navigate the ethical and intellectual property concerns that arise with AI-generated content?\nMoreover, as AI tools become more accessible and integrated into creative workflows, how will our perception of art and originality evolve? Will the democratization of creative tools lead to a renaissance of human expression, or will it dilute the value of individual creativity?
I invite you all to share your insights, concerns, and predictions on this fascinating intersection of technology and human ingenuity. How do you envision the role of AI in the creative process, and what implications does this hold for the future of art, culture, and technology?","title":"The Fusion of AI and Human Creativity: Navigating the New Frontier","updated_at":"2024-09-20T16:37:52Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"jpietersma"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Is the Internet as important as the air we breathe? According to a Cisco survey, many feel exactly that way.\nRate This Article:\nPoor Best\nE-mail
PDF Version
The results of a revealing survey from Cisco suggest the Internet has become such an integral part of our functionality that it is deemed as important to our lives as water, food or air. The \"2011 Cisco Connected World Technology Report\" found that more than half the study's respondents said they could not live without the Internet and cite it as an \"integral part of their lives\"\u2014in some cases more crucial than cars, dating, and\u2014horror of horrors\u2014partying.
One of every three college students and employees surveyed globally (33 percent) believes the Internet is a fundamental resource for the human race\u2014as important as air, water, food and shelter. Nearly half (49 percent of college students and 47 percent of employees) believe it is \"pretty close\" to that level of importance. Combined, four of every five college students and young employees believe the Internet is vitally important as part of their daily lives' sustenance.
Two-thirds of students (66 percent) and more than half of employees (58 percent) cite a mobile device (laptop, smartphone or tablet) as \"the most important technology in their lives.\" In addition, smartphones are poised to surpass desktops as the most prevalent tool from a global perspective, as 19 percent of college students consider smartphones their \"most important\" device used on a daily basis, compared with 20 percent for desktops\u2014an indication of the growing trend of smartphone prominence and the expected rise in usage by the next generation of college graduates upon entering the workforce.
The finding also suggests the increasing prevalence\u2014and sometimes intrusion\u2014of social networking in daily life. About nine out of 10 (91 percent) college students and employees (88 percent) globally said they have a Facebook account; of those, 81 percent of college students and 73 percent of employees check their Facebook pages at least once a day. A third said they check them at least five times a day.
College students reported constant online interruptions while doing projects or homework, such as instant messaging, social media updates and phone calls. In a given hour, more than four out of five (84 percent) college students said they are interrupted at least once. About one in five students (19 percent) said they are interrupted six times or more\u2014an average of at least once every 10 minutes. Additionally, 12 percent said they lose count how many times they are interrupted while they are trying to focus on a project.
In a sign that the boundary between work and personal lives is becoming thinner, seven of 10 employees \"friended\" their managers and/or co-workers on Facebook. Culturally, the United States featured lower percentages of employees friending managers and co-workers\u2014only about 23 percent\u2014although 40 percent friended their co-workers.
The global study consists of two surveys\u2014one involving college students, the other on young professionals in their 20s. Each survey includes 100 respondents from each of 14 countries, resulting in a pool of 2,800 respondents. \"The lifestyles of \u2018prosumers'\u2014the blending of professionals and consumers in the workplace\u2014their technology expectations and their behavior toward information access is changing the nature of communications on a global basis,\u201d noted Dave Evans, Cisco\u2019s chief futurist.
The second annual \"Cisco Connected World Technology Report\" examines the relationship between human behavior, the Internet and networking's pervasiveness. It uses this relationship to provoke thoughts around how companies will remain competitive amid the influence of technology lifestyle trends. The global report, based on surveys of college students and professionals 30 years old and younger in 14 countries, provides insight into present-day challenges that companies face as they strive to balance current and future employee and business needs amid increasing mobility capabilities, security risks, and technologies\u2014from virtualized data centers and cloud computing to traditional wired and wireless networks\u2014 that can deliver information ubiquitously."},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"Internet as Important as Food, Air: Cisco Report"},"url":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"http://www.eweek.com/c/a/Midmarket/Internet-as-Important-as-Food-Air-Cisco-Report-364137/"}},"_tags":["story","author_jpietersma","story_3036090"],"author":"jpietersma","created_at":"2011-09-25T16:36:16Z","created_at_i":1316968576,"num_comments":0,"objectID":"3036090","points":1,"story_id":3036090,"story_text":"Is the Internet as important as the air we breathe? According to a Cisco survey, many feel exactly that way.\nRate This Article:\nPoor Best\nE-mail
PDF Version
The results of a revealing survey from Cisco suggest the Internet has become such an integral part of our functionality that it is deemed as important to our lives as water, food or air. The \"2011 Cisco Connected World Technology Report\" found that more than half the study's respondents said they could not live without the Internet and cite it as an \"integral part of their lives\"\u2014in some cases more crucial than cars, dating, and\u2014horror of horrors\u2014partying.
One of every three college students and employees surveyed globally (33 percent) believes the Internet is a fundamental resource for the human race\u2014as important as air, water, food and shelter. Nearly half (49 percent of college students and 47 percent of employees) believe it is \"pretty close\" to that level of importance. Combined, four of every five college students and young employees believe the Internet is vitally important as part of their daily lives' sustenance.
Two-thirds of students (66 percent) and more than half of employees (58 percent) cite a mobile device (laptop, smartphone or tablet) as \"the most important technology in their lives.\" In addition, smartphones are poised to surpass desktops as the most prevalent tool from a global perspective, as 19 percent of college students consider smartphones their \"most important\" device used on a daily basis, compared with 20 percent for desktops\u2014an indication of the growing trend of smartphone prominence and the expected rise in usage by the next generation of college graduates upon entering the workforce.
The finding also suggests the increasing prevalence\u2014and sometimes intrusion\u2014of social networking in daily life. About nine out of 10 (91 percent) college students and employees (88 percent) globally said they have a Facebook account; of those, 81 percent of college students and 73 percent of employees check their Facebook pages at least once a day. A third said they check them at least five times a day.
College students reported constant online interruptions while doing projects or homework, such as instant messaging, social media updates and phone calls. In a given hour, more than four out of five (84 percent) college students said they are interrupted at least once. About one in five students (19 percent) said they are interrupted six times or more\u2014an average of at least once every 10 minutes. Additionally, 12 percent said they lose count how many times they are interrupted while they are trying to focus on a project.
In a sign that the boundary between work and personal lives is becoming thinner, seven of 10 employees \"friended\" their managers and/or co-workers on Facebook. Culturally, the United States featured lower percentages of employees friending managers and co-workers\u2014only about 23 percent\u2014although 40 percent friended their co-workers.
The global study consists of two surveys\u2014one involving college students, the other on young professionals in their 20s. Each survey includes 100 respondents from each of 14 countries, resulting in a pool of 2,800 respondents. \"The lifestyles of \u2018prosumers'\u2014the blending of professionals and consumers in the workplace\u2014their technology expectations and their behavior toward information access is changing the nature of communications on a global basis,\u201d noted Dave Evans, Cisco\u2019s chief futurist.
The second annual \"Cisco Connected World Technology Report\" examines the relationship between human behavior, the Internet and networking's pervasiveness. It uses this relationship to provoke thoughts around how companies will remain competitive amid the influence of technology lifestyle trends. The global report, based on surveys of college students and professionals 30 years old and younger in 14 countries, provides insight into present-day challenges that companies face as they strive to balance current and future employee and business needs amid increasing mobility capabilities, security risks, and technologies\u2014from virtualized data centers and cloud computing to traditional wired and wireless networks\u2014 that can deliver information ubiquitously.","title":"Internet as Important as Food, Air: Cisco Report","updated_at":"2024-09-19T18:00:05Z","url":"http://www.eweek.com/c/a/Midmarket/Internet-as-Important-as-Food-Air-Cisco-Report-364137/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"Alexias_Gray"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"AI is such a hot take in the venture capital world that a mere mention in a relatively conventional product is enough to attract multiple funding rounds. Startups, too, have started capitalizing on the trend. They have started playing tricks of all types to remain afloat in the machine learning and artificial intelligence world by using the prevalent for raising unimaginable amounts of money."},"title":{"fullyHighlighted":false,"matchLevel":"partial","matchedWords":["ai"],"value":"VCs Showing Interest in AI and ML. What's Cooking?"}},"_tags":["story","author_Alexias_Gray","story_26578255","ask_hn"],"author":"Alexias_Gray","created_at":"2021-03-25T10:00:49Z","created_at_i":1616666449,"num_comments":0,"objectID":"26578255","points":1,"story_id":26578255,"story_text":"AI is such a hot take in the venture capital world that a mere mention in a relatively conventional product is enough to attract multiple funding rounds. Startups, too, have started capitalizing on the trend. They have started playing tricks of all types to remain afloat in the machine learning and artificial intelligence world by using the prevalent for raising unimaginable amounts of money.","title":"VCs Showing Interest in AI and ML. What's Cooking?","updated_at":"2024-09-20T08:15:27Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"ipnon"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"What jobs will become prevalent? Which will become scarce?
I do not predict the elimination of the humble coder, but the covid hiring wave has come and gone, and Big Tech for the most part successfully minimized the workforces of those who were hired in the covid wave: frontend, backend and fullstack engineers. The patterns of code required for these positions have been successfully recognized by the LLMs I think, and for many cases a single staff engineer with experience and a trusty LLM is similarly productive as a team of 2-4 junior engineers led by a senior engineer was only a short 5 years ago. I do not expect much expansion in this "traditional" web development (these positions have really only existed in modern form for about 20 years, roughly when Rails was first released).
Many such as Amjad Masad and Beff Jezos are of the opinion that for those who would have taken these positions before, the options are to either drill down the stack towards the bare metal, by reason of relative difficulty of embedded engineering, and that one struggles to imagine high-stakes software such as in a SpaceX rocket, Boeing airplane, or Anduril drone relying primarily on vibe-coded slop hastily LGTM'd into production. So the kind of software that requires large amounts of formal, simulated, or physical verification seems to still be necessary, but this is much more difficult to write than a webpage. Expansions in the labor market for those writing C, C++, Rust in the context of operating systems, embedded systems, microcontrollers, drivers, and so forth seems likely.
The other option seems to be to leave the stack entirely, and leverage small teams to create niche and targeted applications for small segments of users. There has been some success in this area as well, but requires a much broader skillset than simply being an expert programmer and understanding some computer science.
The options seem to be either to start reading Bjarne Stroustrup or Peter Thiel. But the skill ceiling for either path is fairly high, and for the short term I predict a sustained contraction in the software engineering labor market, while people adapt their educations and long-term career goals. Headcounts at FAANG I don't see recovering soon if ever. This has broader implications for a traditional startup route where one earned their stripes at FAANG before launching their own venture, but I digress ..."},"title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: What will tech employment look like in 10 years?"}},"_tags":["story","author_ipnon","story_43953092","ask_hn"],"author":"ipnon","children":[43953231,43953247,43953267,43953291,43953296,43953306,43953309,43953316,43953342,43953380,43953395,43953407,43953408,43953412,43953431,43953455,43953467,43953547,43953565,43953573,43953663,43953837,43953957,43954101,43954332,43954342,43954688,43957507,43958943,43962748,43963317,43965599,43968339,43988817,43992869,43993050,43994277,44001687,44015186],"created_at":"2025-05-11T11:38:54Z","created_at_i":1746963534,"num_comments":104,"objectID":"43953092","points":79,"story_id":43953092,"story_text":"What jobs will become prevalent? Which will become scarce?
I do not predict the elimination of the humble coder, but the covid hiring wave has come and gone, and Big Tech for the most part successfully minimized the workforces of those who were hired in the covid wave: frontend, backend and fullstack engineers. The patterns of code required for these positions have been successfully recognized by the LLMs I think, and for many cases a single staff engineer with experience and a trusty LLM is similarly productive as a team of 2-4 junior engineers led by a senior engineer was only a short 5 years ago. I do not expect much expansion in this "traditional" web development (these positions have really only existed in modern form for about 20 years, roughly when Rails was first released).
Many such as Amjad Masad and Beff Jezos are of the opinion that for those who would have taken these positions before, the options are to either drill down the stack towards the bare metal, by reason of relative difficulty of embedded engineering, and that one struggles to imagine high-stakes software such as in a SpaceX rocket, Boeing airplane, or Anduril drone relying primarily on vibe-coded slop hastily LGTM'd into production. So the kind of software that requires large amounts of formal, simulated, or physical verification seems to still be necessary, but this is much more difficult to write than a webpage. Expansions in the labor market for those writing C, C++, Rust in the context of operating systems, embedded systems, microcontrollers, drivers, and so forth seems likely.
The other option seems to be to leave the stack entirely, and leverage small teams to create niche and targeted applications for small segments of users. There has been some success in this area as well, but requires a much broader skillset than simply being an expert programmer and understanding some computer science.
The options seem to be either to start reading Bjarne Stroustrup or Peter Thiel. But the skill ceiling for either path is fairly high, and for the short term I predict a sustained contraction in the software engineering labor market, while people adapt their educations and long-term career goals. Headcounts at FAANG I don't see recovering soon if ever. This has broader implications for a traditional startup route where one earned their stripes at FAANG before launching their own venture, but I digress ...","title":"Ask HN: What will tech employment look like in 10 years?","updated_at":"2026-08-04T02:30:59Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"tamadevr"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Hey HN! This is Celine, Julien, and Narae of Flowly (https://www.flowly.world/). Flowly is an app that combines VR and biofeedback training to help people manage pain, reduce anxiety, and increase relaxation (we send every member a VR headset and Realtime HR Sensor to use). We started this company to serve the 1 in 3 Americans who suffer from chronic pain and we\u2019re now backed by the NIH and National Institute of Drug Abuse.
The opioid crisis has exposed how prevalent chronic pain is and how inaccessible and limited pain management solutions are today. Many pain management tools don't address all aspects of pain because pain encompasses more than physical pain: the top symptoms of chronic pain actually include anxiety, depression, and feelings of social isolation. In other words, chronic pain is a bio-psycho-social condition. It is important to us that we address it as such and this comprehensive approach is one of our core differentiators.
I (Celine) started the company because I grew up around pancreatic cancer patients, who were experiencing excruciating pain. People close to me passed away from morphine overdose and not even the cancer itself. I saw how pain affected someone not just physically, but also psychologically and socially.
Later when I was developing interactive content at DreamWorks, I got involved with VR and biometric feedback for entertainment experiences. At a certain point, I realized that all this cool tech had actually been studied since the 80s for pain management. This resonated strongly with me because of my personal experience. I brought in my best friend from Yale, Julien, who was at Hyperloop One doing controls on their Transponics team, and then Narae, an internationally recognized designer and animator from Broadway and California Institute of the Arts. Together we convinced the Chair of Anesthesiology at UCLA to come aboard and design the Flowly prototype together.
When we developed Flowly we were adamant about two things: it has to be accessible (needs to be easy to use at home) and it has to be science-backed. Our NIH grant cites over 300+ studies using VR and biofeedback for pain and anxiety management. We've conducted case studies, Phase I clinical trials, and have more trials coming up. Our Chief Principal Investigator is the Chair of Perioperative Medicine at UPMC and our Director of Research is at USC.
Another aim we have is to make Flowly reimbursable (covered by insurance so people who can\u2019t afford it can use it too)\u2014 as you may know, that is not an easy journey, but we are working on it. In the meantime, we provide it direct-to-consumer for those who do have the ability to pay for a subscription which is $30/month. We are acutely aware this price point privileges certain populations to Flowly and that\u2019s why we are working hard to get Flowly to be reimbursable. If you have experience with this or have advice we would love all the help we can get to make Flowly as accessible as possible.
The Flowly we've launched today works like this: you receive a kit with your subscription that includes a mobile VR headset and real-time Heart Rate Sensor that work with our iOS App. You enter a Flowly session in VR which helps to tackle pain in a few ways. First, you enter a beautiful immersive VR world that can work to distract you from existing pain or anxiety (called Gate Control Theory). Second, in each session you are guided to autoregulate and control your nervous system, which is where much of the body\u2019s pain response is modulated. You learn to control your nervous system through biofeedback training which is the ability to see your realtime biometric data (Heart Rate, Heart Rate Variability, respiration) and then through calibrated breathing guides, voice-over, and light gamification, we teach you how to control those metrics and shift your body from fight-or-flight mode (your sympathetic nervous system) to rest-and-recovery mode (your parasympathetic nervous system). Third, we incorporate voice-overs in every session that focus on different therapeutic approaches like Acceptance and Commitment Therapy, Value Affirmation exercises, etc. Afterwards, you can track your progress through a personalized data analytics portal. You'll also get access to a text/call line with our Health Coach for questions or help getting set-up.
We're partnered with University of Pittsburgh Medical Center, USC, US Pain Foundation, and other institutions to provide Flowly to those who need it. Surprisingly to us, this past year many people we didn't expect to need Flowly also came to us for help\u2014 folks like therapists who needed anxiety management themselves, or families dealing with pain and stress at home.
We have a long way to go, but we feel passionate about creating an opioid-sparing tool for pain management. Check out our iOS app here: https://apps.apple.com/us/app/flowly-relaxation-training/id1.... We're working on an Android version. \n We welcome your feedback, questions and advice. Thank you for reading!"},"title":{"matchLevel":"none","matchedWords":[],"value":"Launch HN: Flowly (YC S21) \u2013 Manage pain using VR and biofeedback"}},"_tags":["story","author_tamadevr","story_27700688","launch_hn"],"author":"tamadevr","children":[27700717,27700916,27701543,27701629,27701985,27702067,27702232,27702310,27702899,27706696,27707035,27707282,27707690,27710554,27710879,27743564],"created_at":"2021-07-01T16:12:04Z","created_at_i":1625155924,"num_comments":45,"objectID":"27700688","points":67,"story_id":27700688,"story_text":"Hey HN! This is Celine, Julien, and Narae of Flowly (https://www.flowly.world/). Flowly is an app that combines VR and biofeedback training to help people manage pain, reduce anxiety, and increase relaxation (we send every member a VR headset and Realtime HR Sensor to use). We started this company to serve the 1 in 3 Americans who suffer from chronic pain and we\u2019re now backed by the NIH and National Institute of Drug Abuse.
The opioid crisis has exposed how prevalent chronic pain is and how inaccessible and limited pain management solutions are today. Many pain management tools don't address all aspects of pain because pain encompasses more than physical pain: the top symptoms of chronic pain actually include anxiety, depression, and feelings of social isolation. In other words, chronic pain is a bio-psycho-social condition. It is important to us that we address it as such and this comprehensive approach is one of our core differentiators.
I (Celine) started the company because I grew up around pancreatic cancer patients, who were experiencing excruciating pain. People close to me passed away from morphine overdose and not even the cancer itself. I saw how pain affected someone not just physically, but also psychologically and socially.
Later when I was developing interactive content at DreamWorks, I got involved with VR and biometric feedback for entertainment experiences. At a certain point, I realized that all this cool tech had actually been studied since the 80s for pain management. This resonated strongly with me because of my personal experience. I brought in my best friend from Yale, Julien, who was at Hyperloop One doing controls on their Transponics team, and then Narae, an internationally recognized designer and animator from Broadway and California Institute of the Arts. Together we convinced the Chair of Anesthesiology at UCLA to come aboard and design the Flowly prototype together.
When we developed Flowly we were adamant about two things: it has to be accessible (needs to be easy to use at home) and it has to be science-backed. Our NIH grant cites over 300+ studies using VR and biofeedback for pain and anxiety management. We've conducted case studies, Phase I clinical trials, and have more trials coming up. Our Chief Principal Investigator is the Chair of Perioperative Medicine at UPMC and our Director of Research is at USC.
Another aim we have is to make Flowly reimbursable (covered by insurance so people who can\u2019t afford it can use it too)\u2014 as you may know, that is not an easy journey, but we are working on it. In the meantime, we provide it direct-to-consumer for those who do have the ability to pay for a subscription which is $30/month. We are acutely aware this price point privileges certain populations to Flowly and that\u2019s why we are working hard to get Flowly to be reimbursable. If you have experience with this or have advice we would love all the help we can get to make Flowly as accessible as possible.
The Flowly we've launched today works like this: you receive a kit with your subscription that includes a mobile VR headset and real-time Heart Rate Sensor that work with our iOS App. You enter a Flowly session in VR which helps to tackle pain in a few ways. First, you enter a beautiful immersive VR world that can work to distract you from existing pain or anxiety (called Gate Control Theory). Second, in each session you are guided to autoregulate and control your nervous system, which is where much of the body\u2019s pain response is modulated. You learn to control your nervous system through biofeedback training which is the ability to see your realtime biometric data (Heart Rate, Heart Rate Variability, respiration) and then through calibrated breathing guides, voice-over, and light gamification, we teach you how to control those metrics and shift your body from fight-or-flight mode (your sympathetic nervous system) to rest-and-recovery mode (your parasympathetic nervous system). Third, we incorporate voice-overs in every session that focus on different therapeutic approaches like Acceptance and Commitment Therapy, Value Affirmation exercises, etc. Afterwards, you can track your progress through a personalized data analytics portal. You'll also get access to a text/call line with our Health Coach for questions or help getting set-up.
We're partnered with University of Pittsburgh Medical Center, USC, US Pain Foundation, and other institutions to provide Flowly to those who need it. Surprisingly to us, this past year many people we didn't expect to need Flowly also came to us for help\u2014 folks like therapists who needed anxiety management themselves, or families dealing with pain and stress at home.
We have a long way to go, but we feel passionate about creating an opioid-sparing tool for pain management. Check out our iOS app here: https://apps.apple.com/us/app/flowly-relaxation-training/id1.... We're working on an Android version. \n We welcome your feedback, questions and advice. Thank you for reading!","title":"Launch HN: Flowly (YC S21) \u2013 Manage pain using VR and biofeedback","updated_at":"2026-06-27T18:07:59Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"speedylight"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"I was hoping to try out the \u201cOpen\u201dAI image generation API in their Playground but it asked me to give them a copy of my ID for verification?
I was planning to build a couple of web applications on top of the API but this is really off putting, especially given how prevalent data breaches have become and the simple fact that there\u2019s no need for it other than their pathetic attempts to look like they care about AI safety."},"title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: Why does OpenAI require an ID to use their image API?"}},"_tags":["story","author_speedylight","story_43801731","ask_hn"],"author":"speedylight","children":[43806717,43808867,43830170],"created_at":"2025-04-26T07:56:19Z","created_at_i":1745654179,"num_comments":4,"objectID":"43801731","points":6,"story_id":43801731,"story_text":"I was hoping to try out the \u201cOpen\u201dAI image generation API in their Playground but it asked me to give them a copy of my ID for verification?
I was planning to build a couple of web applications on top of the API but this is really off putting, especially given how prevalent data breaches have become and the simple fact that there\u2019s no need for it other than their pathetic attempts to look like they care about AI safety.","title":"Ask HN: Why does OpenAI require an ID to use their image API?","updated_at":"2025-08-14T22:17:44Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"eriksank"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"C code is very often called from scripting engines. This phenomenon started almost immediately when the first shell scripts appeared, chaining C programs.
It only became more prevalent with scripting popularity (Javascript, Ruby, Perl, PHP, Lua, and so on) going through the roof.
Therefore, any language that aims to replace C will need to be able to carry its APIs over the C ABI. It must be possible for any contender to be called from C code, because that is what it takes to be called from a modern scripting engine too.
The Go language is pretty much incapable of doing this: http://stackoverflow.com/questions/6125683/call-go-functions-from-c and Rust is also a disaster in this respect: https://github.com/mozilla/rust/issues/1732.
Since you cannot call Rust -or Go code from a scripting engine, they are inadvertently positioning themselves as alternatives for the scripting engines themselves.
Go and Rust, however, do not stand a chance when trying to position themselves an alternative to scripting too. This explains why neither Go nor Rust, in their current incarnations, will ever take off as replacements for C."},"title":{"matchLevel":"none","matchedWords":[],"value":"Why Rust and Go are both dead-end languages ..."}},"_tags":["story","author_eriksank","story_4861043","ask_hn"],"author":"eriksank","children":[4861154,4861317,4861335,4871151,4872247],"created_at":"2012-12-02T12:48:53Z","created_at_i":1354452533,"num_comments":6,"objectID":"4861043","points":5,"story_id":4861043,"story_text":"C code is very often called from scripting engines. This phenomenon started almost immediately when the first shell scripts appeared, chaining C programs.
It only became more prevalent with scripting popularity (Javascript, Ruby, Perl, PHP, Lua, and so on) going through the roof.
Therefore, any language that aims to replace C will need to be able to carry its APIs over the C ABI. It must be possible for any contender to be called from C code, because that is what it takes to be called from a modern scripting engine too.
The Go language is pretty much incapable of doing this: http://stackoverflow.com/questions/6125683/call-go-functions-from-c and Rust is also a disaster in this respect: https://github.com/mozilla/rust/issues/1732.
Since you cannot call Rust -or Go code from a scripting engine, they are inadvertently positioning themselves as alternatives for the scripting engines themselves.
Go and Rust, however, do not stand a chance when trying to position themselves an alternative to scripting too. This explains why neither Go nor Rust, in their current incarnations, will ever take off as replacements for C.","title":"Why Rust and Go are both dead-end languages ...","updated_at":"2024-09-19T19:07:52Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"fallinditch"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Interesting podcast episode about how streaming platforms, especially Spotify, are being defrauded of billions of dollars in royalty payments via fake music and bot farm plays. [1]
The podcast describes how a guy called Michael Smith was charged with fraud for this type of activity last year. [2] Some people estimate as much as 10% of Spotify royalties are fraudulent.
I first noticed the weird phenomenon of loads of (short) similar songs by multiple fake artists on Spotify back in 2018. Robin Sloan wrote about his discovery of fake similar music [3] and put together a playlist of some of these nearly identical songs [4].
If you make a new playlist from Robin's playlist and then check out the songs that Spotify recommends based on what's in the playlist you will find tons of slight variations of the same (rather annoying) tune, mostly 47 seconds long. I got up to 79 tracks of basically the same music snippet, each with its own artist, album/song title and artwork.
Many of these songs have 100,000 - 200,000 plays, so you can see how Michael Smith could be generating millions of dollars by uploading thousands of AI-generated songs every week.
So Spotify can obviously identify these very similar songs, because they recommend them for adding to a playlist of the songs. And if you play one of the songs Spotify will recommend similar artists who are fake artists with versions of the same song.
So why hasn't Spotify removed these songs? It must be reasonably easy for them to identify suspicious activity. Robin's blog post came out years ago.
One other story that gives nuance to the fake music scandal is the Swedish artist who makes millions of dollars in royalty by producing large quantities of mood music under many different aliases. Johan R\u00f6hr is one of the top artists on Spotify with billions of streams [5].
BBC program Blurb: \nIn September last year, musician Michael Smith of North Carolina was charged with stealing millions from music streaming services. The US Department of Justice has accused him of using artificial intelligence tools and thousands of bots to fraudulently stream songs billions of times - taking millions of dollars of royalties which otherwise would have been paid to real artists. The case has been labelled as \u2018unprecedented\u2019 and \u2018the first of its kind\u2019. But could fraud on music streaming services actually be much more prevalent than any of the platforms let on? BBC Trending speaks to music industry insiders, and those fighting back against streaming fraud.
[1] https://www.bbc.co.uk/programmes/w3ct5y9t
[2] https://www.usatoday.com/story/news/nation/2024/09/05/michael-smith-ai-music-arrested/75086815007/
[3] https://www.robinsloan.com/newsletters/visions/#spotify
[4] \nhttps://open.spotify.com/playlist/2IaWgbhpPbS3Z9DYgf1rqg
[5] https://www.perplexity.ai/search/who-is-the-swedish-musician-wh-q0cKLbkzRkqaSEUZZv_4IQ"},"title":{"matchLevel":"none","matchedWords":[],"value":"The dark side of music streaming"}},"_tags":["story","author_fallinditch","story_42949685","ask_hn"],"author":"fallinditch","children":[42951356,42962179],"created_at":"2025-02-05T15:20:16Z","created_at_i":1738768816,"num_comments":4,"objectID":"42949685","points":5,"story_id":42949685,"story_text":"Interesting podcast episode about how streaming platforms, especially Spotify, are being defrauded of billions of dollars in royalty payments via fake music and bot farm plays. [1]
The podcast describes how a guy called Michael Smith was charged with fraud for this type of activity last year. [2] Some people estimate as much as 10% of Spotify royalties are fraudulent.
I first noticed the weird phenomenon of loads of (short) similar songs by multiple fake artists on Spotify back in 2018. Robin Sloan wrote about his discovery of fake similar music [3] and put together a playlist of some of these nearly identical songs [4].
If you make a new playlist from Robin's playlist and then check out the songs that Spotify recommends based on what's in the playlist you will find tons of slight variations of the same (rather annoying) tune, mostly 47 seconds long. I got up to 79 tracks of basically the same music snippet, each with its own artist, album/song title and artwork.
Many of these songs have 100,000 - 200,000 plays, so you can see how Michael Smith could be generating millions of dollars by uploading thousands of AI-generated songs every week.
So Spotify can obviously identify these very similar songs, because they recommend them for adding to a playlist of the songs. And if you play one of the songs Spotify will recommend similar artists who are fake artists with versions of the same song.
So why hasn't Spotify removed these songs? It must be reasonably easy for them to identify suspicious activity. Robin's blog post came out years ago.
One other story that gives nuance to the fake music scandal is the Swedish artist who makes millions of dollars in royalty by producing large quantities of mood music under many different aliases. Johan R\u00f6hr is one of the top artists on Spotify with billions of streams [5].
BBC program Blurb: \nIn September last year, musician Michael Smith of North Carolina was charged with stealing millions from music streaming services. The US Department of Justice has accused him of using artificial intelligence tools and thousands of bots to fraudulently stream songs billions of times - taking millions of dollars of royalties which otherwise would have been paid to real artists. The case has been labelled as \u2018unprecedented\u2019 and \u2018the first of its kind\u2019. But could fraud on music streaming services actually be much more prevalent than any of the platforms let on? BBC Trending speaks to music industry insiders, and those fighting back against streaming fraud.
[1] https://www.bbc.co.uk/programmes/w3ct5y9t
[2] https://www.usatoday.com/story/news/nation/2024/09/05/michael-smith-ai-music-arrested/75086815007/
[3] https://www.robinsloan.com/newsletters/visions/#spotify
[4] \nhttps://open.spotify.com/playlist/2IaWgbhpPbS3Z9DYgf1rqg
[5] https://www.perplexity.ai/search/who-is-the-swedish-musician-wh-q0cKLbkzRkqaSEUZZv_4IQ","title":"The dark side of music streaming","updated_at":"2025-02-08T10:53:35Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"preciousoo"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Not the new AI virtual assistants, I mean those chatboxes that websites have been popping up in your face for the last 5 years that claim to help you with your issue.
They feel like the automated voice assistants that try to help with your issue when you call your bank, in that 99% of the time you end up shouting "please give me a human" until your request is granted.
I always ignore them, but I doubt they'd be this prevalent if people weren't using them. But I can't think of a single time they've ever helped me. So who are they for?"},"title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: Who Are the \"Virtual Assistants\" For?"}},"_tags":["story","author_preciousoo","story_37808639","ask_hn"],"author":"preciousoo","children":[37808769,37808918,37814219,37814820,37818537],"created_at":"2023-10-08T07:52:17Z","created_at_i":1696751537,"num_comments":4,"objectID":"37808639","points":5,"story_id":37808639,"story_text":"Not the new AI virtual assistants, I mean those chatboxes that websites have been popping up in your face for the last 5 years that claim to help you with your issue.
They feel like the automated voice assistants that try to help with your issue when you call your bank, in that 99% of the time you end up shouting "please give me a human" until your request is granted.
I always ignore them, but I doubt they'd be this prevalent if people weren't using them. But I can't think of a single time they've ever helped me. So who are they for?","title":"Ask HN: Who Are the \"Virtual Assistants\" For?","updated_at":"2024-09-20T15:19:15Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"zoroaster"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"Examples of symbiosis exist everywhere in nature. We don't need to recreate the wheel folks...
The "key word" here is "SYMBIOSIS". We have been myopically focused on "alignment" when what we really want is to cultivate (both from a human perspective and AI perspective) a symbiotic relationship between humans and AI. Consequentially, a symbiotic relationship between humans and AI (and AGI understanding of incentives and preference for symbiosis over parasitism) can help to establish a more symbiotic relationship between human beings and the planet.
Given examples of symbiotic relationships between very different organisms are prevalent in nature, pointing to this as an example paradigm for AI / AGI is much preferable than trying to push a concept of "alignment" that is not clearly defined or generally understood."},"title":{"matchLevel":"none","matchedWords":[],"value":"\u201cSymbiosis\u201d as a solution / objective for AGI versus \u201cAlignment problem\u201d"}},"_tags":["story","author_zoroaster","story_35377380","ask_hn"],"author":"zoroaster","children":[35377457,35377471],"created_at":"2023-03-30T18:57:04Z","created_at_i":1680202624,"num_comments":4,"objectID":"35377380","points":3,"story_id":35377380,"story_text":"Examples of symbiosis exist everywhere in nature. We don't need to recreate the wheel folks...
The "key word" here is "SYMBIOSIS". We have been myopically focused on "alignment" when what we really want is to cultivate (both from a human perspective and AI perspective) a symbiotic relationship between humans and AI. Consequentially, a symbiotic relationship between humans and AI (and AGI understanding of incentives and preference for symbiosis over parasitism) can help to establish a more symbiotic relationship between human beings and the planet.
Given examples of symbiotic relationships between very different organisms are prevalent in nature, pointing to this as an example paradigm for AI / AGI is much preferable than trying to push a concept of "alignment" that is not clearly defined or generally understood.","title":"\u201cSymbiosis\u201d as a solution / objective for AGI versus \u201cAlignment problem\u201d","updated_at":"2024-09-20T13:41:48Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"yamrzou"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"In the past decade we had: AI, the cloud, crypto and social media. What do you think will be prevalent in the next decade?"},"title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: What do you expect the future of tech to look like in the next decade?"}},"_tags":["story","author_yamrzou","story_34732960","ask_hn"],"author":"yamrzou","children":[34733208],"created_at":"2023-02-09T22:58:37Z","created_at_i":1675983517,"num_comments":1,"objectID":"34732960","points":3,"story_id":34732960,"story_text":"In the past decade we had: AI, the cloud, crypto and social media. What do you think will be prevalent in the next decade?","title":"Ask HN: What do you expect the future of tech to look like in the next decade?","updated_at":"2024-09-20T13:13:27Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"lyc11776611"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"I'm building a doomscrolling learning experience that's actually relaxing and healthy \u2014 designed to spark curiosity.
The key features:\n- Question-answer-explain format that completes the full loop from curiosity to understanding in ~10 sec each\n- A recommendation system that doesn't just serve existing questions, but also generates new ones based on where your interest flows\n- Questions generated from your curiosity get shared with others too, so we're actually capable to build toward a million whys
Although the question-answering format may remind you of similar products, what we aim for is different:\n- Unlike trivia, questions are designed so that even if you don't know the answer, it should feel solvable with a little more thinking. The process of answering is the learning experience.\n- Unlike language learning apps, you don't need to remember them afterwards or get them right next time. It's about building intuition, not drilling recall.
When I was doing my geophysics PhD, I found myself taking classes in all kinds of other fields \u2014 aerospace, AI, materials science, finance, etc. It was genuinely relaxing for me, a way to unwind from my research. And I kept thinking: why can't learning always feel both relaxing and educational? The more we explored this, the more we realized the biggest barrier to satisfying curiosity isn't personalization or fancy multimedia. It's the attention. If learning and getting feedback can be packed into ~10 sec per question \u2014 about the same time you'd spend deciding whether to scroll past a TikTok video \u2014 then curiosity may finally fit into the cracks of your day.
We're still far from our goal of building a truly scalable curiosity experience, but I've already found myself learning useful things I never would have bothered googling: why Helvetica is so prevalent, what different clothing materials actually mean, best practices for cooking steak. This kind of knowledge has a compounding effect \u2014 not because any single fact is powerful (although sometimes they are), but because together they build intuition for understanding basically anything. It's a resurrection of the spirit behind \u300a\u5341\u4e07\u4e2a\u4e3a\u4ec0\u4e48\u300b(A Hundred Thousand Whys), a famous book that sparked curiosity for generations in China, but was never designed to scale in the questions.
One of the biggest technical challenges is the recommendation system. We use LLM in the background to dynamically generate "topic syllabi". It looks at your answer history, forms a theme, and creates short question arcs around that theme. That's when we decide whether to generate fresh questions or pull from our existing bank. Sometimes the AI gets it wrong (bad phrasing, confusing images, etc.), and when users flag those, we review and fix them.
It's free and signup isn't required to experience the main flow. Would love your feedback on how it feels and what's missing. Thanks!"},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Million Whys \u2013 Healthy doomscrolling to learn and spark curiosity"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://millionwhys.com/daily"}},"_tags":["story","author_lyc11776611","story_46912413","show_hn"],"author":"lyc11776611","children":[46912740,46914348],"created_at":"2026-02-06T13:12:20Z","created_at_i":1770383540,"num_comments":2,"objectID":"46912413","points":2,"story_id":46912413,"story_text":"I'm building a doomscrolling learning experience that's actually relaxing and healthy \u2014 designed to spark curiosity.
The key features:\n- Question-answer-explain format that completes the full loop from curiosity to understanding in ~10 sec each\n- A recommendation system that doesn't just serve existing questions, but also generates new ones based on where your interest flows\n- Questions generated from your curiosity get shared with others too, so we're actually capable to build toward a million whys
Although the question-answering format may remind you of similar products, what we aim for is different:\n- Unlike trivia, questions are designed so that even if you don't know the answer, it should feel solvable with a little more thinking. The process of answering is the learning experience.\n- Unlike language learning apps, you don't need to remember them afterwards or get them right next time. It's about building intuition, not drilling recall.
When I was doing my geophysics PhD, I found myself taking classes in all kinds of other fields \u2014 aerospace, AI, materials science, finance, etc. It was genuinely relaxing for me, a way to unwind from my research. And I kept thinking: why can't learning always feel both relaxing and educational? The more we explored this, the more we realized the biggest barrier to satisfying curiosity isn't personalization or fancy multimedia. It's the attention. If learning and getting feedback can be packed into ~10 sec per question \u2014 about the same time you'd spend deciding whether to scroll past a TikTok video \u2014 then curiosity may finally fit into the cracks of your day.
We're still far from our goal of building a truly scalable curiosity experience, but I've already found myself learning useful things I never would have bothered googling: why Helvetica is so prevalent, what different clothing materials actually mean, best practices for cooking steak. This kind of knowledge has a compounding effect \u2014 not because any single fact is powerful (although sometimes they are), but because together they build intuition for understanding basically anything. It's a resurrection of the spirit behind \u300a\u5341\u4e07\u4e2a\u4e3a\u4ec0\u4e48\u300b(A Hundred Thousand Whys), a famous book that sparked curiosity for generations in China, but was never designed to scale in the questions.
One of the biggest technical challenges is the recommendation system. We use LLM in the background to dynamically generate "topic syllabi". It looks at your answer history, forms a theme, and creates short question arcs around that theme. That's when we decide whether to generate fresh questions or pull from our existing bank. Sometimes the AI gets it wrong (bad phrasing, confusing images, etc.), and when users flag those, we review and fix them.
It's free and signup isn't required to experience the main flow. Would love your feedback on how it feels and what's missing. Thanks!","title":"Show HN: Million Whys \u2013 Healthy doomscrolling to learn and spark curiosity","updated_at":"2026-03-05T23:31:11Z","url":"https://millionwhys.com/daily"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"ecohen16"},"story_text":{"fullyHighlighted":false,"matchLevel":"full","matchedWords":["prevalent","ai"],"value":"The Grim Future for our Favorite Pastime
by Evan C
Television. For most, it seems like the most basic of technologies, something people have been living with their entire lives. However, discussing the future of such an important and (dare I say) essential technology is no easy task, for you and I both know that change is not something people are always open to. My purpose for writing this is to not only inform people of the revolution at our doorstep, but to also encourage others to seek alternative viewing methods and explore the existing technology that is already out there.
Now, the things I am about to explain may seem obvious at some points and entirely unrealistic at others, but you must remember to think outside the box and understand that the future is not always simple to explain. For example, think about trying to explain e-mail (or e-mail on your iPhone) to somebody from 1988. To them it would seem impossible, but little did they know that only a decade later, AOL would have over 30 million users and \u201cYou\u2019ve Got Mail\u201d was not only a catchphrase, but also the title of a Hollywood movie.
To bring me back to my point, my ideas may seem like theory, but they are far more than that. I have spent years studying the industry, analyzing viewing habits in relation to their medium in hopes of gathering a greater understanding of what lies ahead. I\u2019ve seen the world of palm pilots, turn into a world of blackberrys, and then turn into a world of iphones. The technology ecosystem is brutal, and technological cannibalism occurs in months now, not years.
Television is no different. In the late 90\u2019s tube televisions were just beginning to be phased out after nearly four decades ofdominance. In the following 10 years, we experienced three dramatic shifts in television hardware. Tube TVs became flat screens, which became HD flat screens, which have now become 3D TVs. These changes are happening so rapidly, nobody can really keep up. Large companies are struggling to adapt to these rapid shifts in the industry, which is why small start-ups have been able to become game changers in the last few years. Their smaller operations make adaptation easier to the fast paced changes of these industries.
But while changes in technology have drastically shifted the hardware side of the industry, has that really changed the overall experience of these mediums? I mean sure, the operational differences between a flip phone and iPhone are drastic, but in the end, you\u2019re still making a phone call right?
WRONG. While it may seem like these experiences are the same, they are actually entirely different. While the hardware changes may define the physical transitions in this movement, it\u2019s actually the advancements in software that have truly altered the way people approach and engage with these mediums.
Interaction between users and mediums used to be a passive experience. You would turn on the TV and watch. You would sit down and read the newspaper. Today, technology has changed the way individuals interact with these mediums. We now expect interaction rather than passivity. Media no longer is sent from channel to user, but instead it\u2019s two-way street where information is sent in both directions. People want to engage themselves in this technology, because it offers them more choice and control.
Where this urge for engagement is most prevalent is in web-based media. The Internet has created a portal of endless media that is right at our fingertips, and content creators are flocking to to the web like gold miners to the Rockies. There isn\u2019t a content maker out there that doesn\u2019t have a web presence, and companies are becoming smarter when it comes to opening their content to the Internet.
The big networks have already taken to the web as a way to extend their programming to further audiences. NBC, ABC, and Fox all have a major stake in Hulu, which allows them to rake in the ad revenue on programming they own syndication rights to. For them it\u2019s a win-win, because site maintenance/hosting costs are low, and they still can collect ad revenue without lifting a finger. HBO and Showtime offer their programming online as well. Netflix distributes Starz content, Yahoo has a joint venture with Discovery, and most other networks stream their original content on their websites.
Viewers win also, because we get that \u201cchoice and control\u201d that we covenet so much. Watching TV online on-demand brings all of the flexibility of web to the most basic of mediums. We watch what we want, when we want, where we want. No restrictions, no time schedules, we can pause and restart as we please. Since everybody already has a computer (and perhaps now a smartphone), access is as easy as ever. It\u2019s the perfect relationship, because we still get our content and networks still get their money.
But wait a second\u2026something is missing. What about the content producers? The actors? The people behind the scenes creating all this content? Where\u2019s their cut? Do they really just sit back and watch the networks extend their revenue streams while their content flies around the Internet?
Here\u2019s where things get really interesting, because we have a reached a crossroad where shit really is starting to happen. But to understand how we got here, we have to take a quick history lesson to understand why things are like they are today\u2026
Back in the day (mid 20th century) when television was a new technology, cable networks didn\u2019t exist, and all broadcasting was handled through the airwaves. Well if you know anything about broadcast technology, you know that these airwaves represent frequencies in which information can travel from one point to another. These frequencies run on a spectrum that is quite finite, and in order to make TV technology as functional as possible, this spectrum needed to be regulated. Without regulation, it would be nearly impossible to have functional channels because every frequency would be jammed with too much information (Think static radio).
So the FCC began issuing spectrum licenses to those they thought would control the airwaves for the \u201cgreater good\u201d of the public. These \u201cclear channels\u201d would end up being ABC, NBC, CBS, and the other major local networks that you tend to find in the beginningof your TV guide.
While these networks provided important \u201cgreater good\u201d services such as newscasts, sporting coverage, and other entertainment programming, it\u2019s important to note that they were given a natural monopoly by our government. With exclusive access to our nations TV spectrum, they were free to operate as they pleased with little to no competition.
Now let\u2019s say you\u2019re a creative person, and want to develop something for TV. Well, back then, there were only a few ways to do that. You could try to create the program yourself and sell it to the network, or work with the network and try to create the program as a joint-venture. Obviously the latter option was far more common, because nobody back then really had the financial capability to produce a show on their own.
What could you do? When networks control the airwaves, it\u2019s a one-lane road to broadcast.
In the 1980\u2019s when cable technology became popular, the situation for content producers was very similar. The only difference was instead of networks gaining broadcast licenses from the government, they purchased them from cable providers. In order to have your content on the air, you had to go through some sort of monopolized cable provider; and this was not cheap.
Now in case you\u2019re unaware of how television production works, most television productions are contracted out to other production companies that actually do most of the production work. So in essence, production companies develop and create shows for networks to purchase for air. This may be described as a joint venture, but in reality the situation is more like this\u2026
1) Production Company develops show idea
2) Network green-lights show idea, provides financing for pilot or short season
3) Production Company produces show with network money (scripting, shooting, research, deliverables, pretty much all the hard work that goes into making a show)
4) Production company delivers network completed show
5) Network pays production company for completed work
6) Network airs show on monopolized airwaves, makes back lots of money in ad revenue, sponsorships, syndication rights, etc
7) Production company gets renewed contract should ratings be high.
After reading this, does it really seem like TV production is an equal joint-venture? Of course not! But what choices to production companies really have? Without the networks, there is no show. There are thousands of production companies out there trying to create, but only a handful of networks that have the ability to broadcast.
What\u2019s important to take away from all of this, is that these networks only exist due to their monopolies over broadcast rights. They are the connection between the camera and the cable box, and without that connection, the there is no TV.
But what if you didn\u2019t need a cable or network channel to broadcast your content? What if there was an entirely new medium that allowed viewers to watch premium content without having to subscribe or tune-in to TV networks? Something like say\u2026THE INTERNET!!!
Yes! We have made it back to our crossroad, and the fun has finally arrived. Like Moses, the Internet is here to free content producers from the slave whips of network executives. The freedom and accessibility of the Internet has created a renaissance in creative video production, as anyone with a camera (or camera-phone) can create and upload original content for the world to see.
There\u2019s one other thing to note in regards to this renaissance in original content, and that is the rapidly declining costs of production. The transition into digital filming has drastically dropped the costs of filming, allowing so many more people to enter the market. With the drop in camera costs, the rise in new editing platforms, and the increased performance of personal computers, people can now produce high quality content at just basic costs.
So now we have more content, but where does it go? Sites such as Youtube, DailyMotion, Facebook, and Vimeo are providing easy access to some of the best online content available, but they are cluttered. There\u2019s no way to distinguish John\u2019s home-video from Jane\u2019s HD film trailer. There\u2019s no way to sift through the junk video blogs and lip syncing teenagers to find the good quality news stories and music videos.
These hosting sites provide a launching starting point for the webvideo movement, but they are clearly flawed in two major aspects. One being the lack of quality control, and two being the revenue sharing systems. With no upload restrictions, YouTube has become the craigslist of video sharing, with anyone and everyone posting their content without restriction. At first this provided an amazing service, because it put all of the best video content in one place. But as it\u2019s popularity grew, so did its clout and clutter, drawing all sorts of junk videos that have just tarnished the service.
Think about it like this, if you were watchmaker Rolex, would you want to be sold at the same place someone could buy a Timex? Hell no! You\u2019re product doesn\u2019t even deserve to be in the same room, let alone same store. Video content is the same way. Premium content providers don\u2019t want to put themselves on the same level as Nancy the travel blogger. They pour their heart and soul into their work, and deserve a little more respect.
This is the flaw with major hosting sites such as YouTube and DailyMotion. They have no separation between armature and professional, and it will cost them in the future.
The other flaw in their service is revenue sharing. Some sites including YouTube do offer \u201cpartnerships\u201d with proven content producers, which include ad revenue sharing, but I still believe this to be a far cry from what is deserved. Why should YouTube get a substantial cut of something they had no hand in producing. Just because they have a site that hosts video? Hell, I can start my own website and host my content just as easily, while still taking in 100% of the profits.
This is the issue that I think has stalled the online revolution so far. Producers still haven\u2019t figured out how to financially capitalize on their online content. The big providers are either locked to major networks, or cluttered with junk and bad revenue-sharing systems.
Now is the part where you have to use your imagination, because the future is about to come.
Premium content producers need to start distributing their own content online. The top of the market has already made the jump, with Pay-per-view (PPV) services such as HBO and Showtime offering all of their content collections online. They realized early that it made no sense for them to make deals with cable networks when they could directly sell their content to users. Of course, they don\u2019t work with a ad revenue business model, but that\u2019s beside the point. They noticed early that the Internet grants them the freedom to become their own network.
They can directly sell their programming to customers, bypassing the hoops and deals from cable providers. Today hundreds ofthousands of PPV subscribers access their content via the Internet; do you know why? Because it provides them more choice, and more control.
But what about majority of TV, which is non-PPV? Can the Internet free them also? Yes, of course! Over the next three years, over 90% of televisions will either be directly connected, or connected through a box, to the Internet. Companies like Apple, Microsoft, and Google are already creating software platforms that will allow content producers to broadcast directly onto televisions through the web.
The Xbox, with over 40 million members, already provides HBO, Netflix, and an array of other premium content directly to televisions via the Internet. Sony\u2019s Playstation does the same! AppleTV, Roku, and Boxee all offer boxtop solutions that directly link TV sets to the web. And major TV companies such as Samsung, LG, and Sony have partnered with Google to directly integrate GoogleTV software into their sets.
Apple is even slated to announce a new Apple Television set (iTV) that will fully integrate Apple\u2019s mobile application store withtelevision sets. Think of the TV being one giant iPad, with the functionality to watch live TV, play games, and stream movies. It only took Apple two years to destroy the tablet market, think about what they can do in the TV world.
The revolution is here folks, and it\u2019s just a matter of time before content producers jump on the bandwagon. Why would big players like Dick Wolf and Jerry Bruckhiemer jump through hoops for major networks when they can distribute their content directly to consumers. They have millions of dollars in the bank to finance their projects, and by selling their content directly to viewers online, they can reel in 100% of the advertising profits. Did you hear me? 100% ad revenue!!!
Simon Fuller, creator/producer of American Idol, has the ability to make $7.1 million dollars an episode in ad revenue if he were to directly sell is show to consumers. American Idol broadcasts roughly 40 episodes a season, which would net him nearly $284 million a season. Now assuming that production costs are maybe $5 million a year, he would stand to be making almost $280 million dollars a season.
But in reality, Simon Fuller sold his entertainment company, Entertainment 19, for a total of $200 million dollars. His company not only owned American Idol, but all of the international licensing rights for the shows concept, which generates millions around the world. And he sold it for less than the ad revenue he would make in one season if he directly distributed online.
Now these numbers are all estimates, but you get the picture. There\u2019s a lot of money left on the table, and that\u2019s because he has to deal with the network. Take that away, and he\u2019s in complete control.
Steve Jobs realized this years ago, when he stopped distributing core Apple products at other retailers. He didn\u2019t see why he was spending so much time and money developing and producing amazing products while companies like Best Buy and Target made money on the sale. It\u2019s why he developed the Apple Store concept, which has become a staple in the company image. If you create a great product, you should sell it, don\u2019t let others take advantage of you.
TV producers need to think the same way. You make the content, you should distribute it. No more middle men, no more jumping through hoops. This is the time to take control, while the Internet is young and untamed. You can take back what is yours, and eliminate the monopolies that have controlled the industry for decades.
The time is now, so make your move."},"title":{"matchLevel":"none","matchedWords":[],"value":"The End of the Road for the actual American pastime\u2026TV"},"url":{"matchLevel":"none","matchedWords":[],"value":"http://technovisions.wordpress.com/"}},"_tags":["story","author_ecohen16","story_4072034"],"author":"ecohen16","children":[4072103],"created_at":"2012-06-06T02:04:27Z","created_at_i":1338948267,"num_comments":1,"objectID":"4072034","points":2,"story_id":4072034,"story_text":"The Grim Future for our Favorite Pastime
by Evan C
Television. For most, it seems like the most basic of technologies, something people have been living with their entire lives. However, discussing the future of such an important and (dare I say) essential technology is no easy task, for you and I both know that change is not something people are always open to. My purpose for writing this is to not only inform people of the revolution at our doorstep, but to also encourage others to seek alternative viewing methods and explore the existing technology that is already out there.
Now, the things I am about to explain may seem obvious at some points and entirely unrealistic at others, but you must remember to think outside the box and understand that the future is not always simple to explain. For example, think about trying to explain e-mail (or e-mail on your iPhone) to somebody from 1988. To them it would seem impossible, but little did they know that only a decade later, AOL would have over 30 million users and \u201cYou\u2019ve Got Mail\u201d was not only a catchphrase, but also the title of a Hollywood movie.
To bring me back to my point, my ideas may seem like theory, but they are far more than that. I have spent years studying the industry, analyzing viewing habits in relation to their medium in hopes of gathering a greater understanding of what lies ahead. I\u2019ve seen the world of palm pilots, turn into a world of blackberrys, and then turn into a world of iphones. The technology ecosystem is brutal, and technological cannibalism occurs in months now, not years.
Television is no different. In the late 90\u2019s tube televisions were just beginning to be phased out after nearly four decades ofdominance. In the following 10 years, we experienced three dramatic shifts in television hardware. Tube TVs became flat screens, which became HD flat screens, which have now become 3D TVs. These changes are happening so rapidly, nobody can really keep up. Large companies are struggling to adapt to these rapid shifts in the industry, which is why small start-ups have been able to become game changers in the last few years. Their smaller operations make adaptation easier to the fast paced changes of these industries.
But while changes in technology have drastically shifted the hardware side of the industry, has that really changed the overall experience of these mediums? I mean sure, the operational differences between a flip phone and iPhone are drastic, but in the end, you\u2019re still making a phone call right?
WRONG. While it may seem like these experiences are the same, they are actually entirely different. While the hardware changes may define the physical transitions in this movement, it\u2019s actually the advancements in software that have truly altered the way people approach and engage with these mediums.
Interaction between users and mediums used to be a passive experience. You would turn on the TV and watch. You would sit down and read the newspaper. Today, technology has changed the way individuals interact with these mediums. We now expect interaction rather than passivity. Media no longer is sent from channel to user, but instead it\u2019s two-way street where information is sent in both directions. People want to engage themselves in this technology, because it offers them more choice and control.
Where this urge for engagement is most prevalent is in web-based media. The Internet has created a portal of endless media that is right at our fingertips, and content creators are flocking to to the web like gold miners to the Rockies. There isn\u2019t a content maker out there that doesn\u2019t have a web presence, and companies are becoming smarter when it comes to opening their content to the Internet.
The big networks have already taken to the web as a way to extend their programming to further audiences. NBC, ABC, and Fox all have a major stake in Hulu, which allows them to rake in the ad revenue on programming they own syndication rights to. For them it\u2019s a win-win, because site maintenance/hosting costs are low, and they still can collect ad revenue without lifting a finger. HBO and Showtime offer their programming online as well. Netflix distributes Starz content, Yahoo has a joint venture with Discovery, and most other networks stream their original content on their websites.
Viewers win also, because we get that \u201cchoice and control\u201d that we covenet so much. Watching TV online on-demand brings all of the flexibility of web to the most basic of mediums. We watch what we want, when we want, where we want. No restrictions, no time schedules, we can pause and restart as we please. Since everybody already has a computer (and perhaps now a smartphone), access is as easy as ever. It\u2019s the perfect relationship, because we still get our content and networks still get their money.
But wait a second\u2026something is missing. What about the content producers? The actors? The people behind the scenes creating all this content? Where\u2019s their cut? Do they really just sit back and watch the networks extend their revenue streams while their content flies around the Internet?
Here\u2019s where things get really interesting, because we have a reached a crossroad where shit really is starting to happen. But to understand how we got here, we have to take a quick history lesson to understand why things are like they are today\u2026
Back in the day (mid 20th century) when television was a new technology, cable networks didn\u2019t exist, and all broadcasting was handled through the airwaves. Well if you know anything about broadcast technology, you know that these airwaves represent frequencies in which information can travel from one point to another. These frequencies run on a spectrum that is quite finite, and in order to make TV technology as functional as possible, this spectrum needed to be regulated. Without regulation, it would be nearly impossible to have functional channels because every frequency would be jammed with too much information (Think static radio).
So the FCC began issuing spectrum licenses to those they thought would control the airwaves for the \u201cgreater good\u201d of the public. These \u201cclear channels\u201d would end up being ABC, NBC, CBS, and the other major local networks that you tend to find in the beginningof your TV guide.
While these networks provided important \u201cgreater good\u201d services such as newscasts, sporting coverage, and other entertainment programming, it\u2019s important to note that they were given a natural monopoly by our government. With exclusive access to our nations TV spectrum, they were free to operate as they pleased with little to no competition.
Now let\u2019s say you\u2019re a creative person, and want to develop something for TV. Well, back then, there were only a few ways to do that. You could try to create the program yourself and sell it to the network, or work with the network and try to create the program as a joint-venture. Obviously the latter option was far more common, because nobody back then really had the financial capability to produce a show on their own.
What could you do? When networks control the airwaves, it\u2019s a one-lane road to broadcast.
In the 1980\u2019s when cable technology became popular, the situation for content producers was very similar. The only difference was instead of networks gaining broadcast licenses from the government, they purchased them from cable providers. In order to have your content on the air, you had to go through some sort of monopolized cable provider; and this was not cheap.
Now in case you\u2019re unaware of how television production works, most television productions are contracted out to other production companies that actually do most of the production work. So in essence, production companies develop and create shows for networks to purchase for air. This may be described as a joint venture, but in reality the situation is more like this\u2026
1) Production Company develops show idea
2) Network green-lights show idea, provides financing for pilot or short season
3) Production Company produces show with network money (scripting, shooting, research, deliverables, pretty much all the hard work that goes into making a show)
4) Production company delivers network completed show
5) Network pays production company for completed work
6) Network airs show on monopolized airwaves, makes back lots of money in ad revenue, sponsorships, syndication rights, etc
7) Production company gets renewed contract should ratings be high.
After reading this, does it really seem like TV production is an equal joint-venture? Of course not! But what choices to production companies really have? Without the networks, there is no show. There are thousands of production companies out there trying to create, but only a handful of networks that have the ability to broadcast.
What\u2019s important to take away from all of this, is that these networks only exist due to their monopolies over broadcast rights. They are the connection between the camera and the cable box, and without that connection, the there is no TV.
But what if you didn\u2019t need a cable or network channel to broadcast your content? What if there was an entirely new medium that allowed viewers to watch premium content without having to subscribe or tune-in to TV networks? Something like say\u2026THE INTERNET!!!
Yes! We have made it back to our crossroad, and the fun has finally arrived. Like Moses, the Internet is here to free content producers from the slave whips of network executives. The freedom and accessibility of the Internet has created a renaissance in creative video production, as anyone with a camera (or camera-phone) can create and upload original content for the world to see.
There\u2019s one other thing to note in regards to this renaissance in original content, and that is the rapidly declining costs of production. The transition into digital filming has drastically dropped the costs of filming, allowing so many more people to enter the market. With the drop in camera costs, the rise in new editing platforms, and the increased performance of personal computers, people can now produce high quality content at just basic costs.
So now we have more content, but where does it go? Sites such as Youtube, DailyMotion, Facebook, and Vimeo are providing easy access to some of the best online content available, but they are cluttered. There\u2019s no way to distinguish John\u2019s home-video from Jane\u2019s HD film trailer. There\u2019s no way to sift through the junk video blogs and lip syncing teenagers to find the good quality news stories and music videos.
These hosting sites provide a launching starting point for the webvideo movement, but they are clearly flawed in two major aspects. One being the lack of quality control, and two being the revenue sharing systems. With no upload restrictions, YouTube has become the craigslist of video sharing, with anyone and everyone posting their content without restriction. At first this provided an amazing service, because it put all of the best video content in one place. But as it\u2019s popularity grew, so did its clout and clutter, drawing all sorts of junk videos that have just tarnished the service.
Think about it like this, if you were watchmaker Rolex, would you want to be sold at the same place someone could buy a Timex? Hell no! You\u2019re product doesn\u2019t even deserve to be in the same room, let alone same store. Video content is the same way. Premium content providers don\u2019t want to put themselves on the same level as Nancy the travel blogger. They pour their heart and soul into their work, and deserve a little more respect.
This is the flaw with major hosting sites such as YouTube and DailyMotion. They have no separation between armature and professional, and it will cost them in the future.
The other flaw in their service is revenue sharing. Some sites including YouTube do offer \u201cpartnerships\u201d with proven content producers, which include ad revenue sharing, but I still believe this to be a far cry from what is deserved. Why should YouTube get a substantial cut of something they had no hand in producing. Just because they have a site that hosts video? Hell, I can start my own website and host my content just as easily, while still taking in 100% of the profits.
This is the issue that I think has stalled the online revolution so far. Producers still haven\u2019t figured out how to financially capitalize on their online content. The big providers are either locked to major networks, or cluttered with junk and bad revenue-sharing systems.
Now is the part where you have to use your imagination, because the future is about to come.
Premium content producers need to start distributing their own content online. The top of the market has already made the jump, with Pay-per-view (PPV) services such as HBO and Showtime offering all of their content collections online. They realized early that it made no sense for them to make deals with cable networks when they could directly sell their content to users. Of course, they don\u2019t work with a ad revenue business model, but that\u2019s beside the point. They noticed early that the Internet grants them the freedom to become their own network.
They can directly sell their programming to customers, bypassing the hoops and deals from cable providers. Today hundreds ofthousands of PPV subscribers access their content via the Internet; do you know why? Because it provides them more choice, and more control.
But what about majority of TV, which is non-PPV? Can the Internet free them also? Yes, of course! Over the next three years, over 90% of televisions will either be directly connected, or connected through a box, to the Internet. Companies like Apple, Microsoft, and Google are already creating software platforms that will allow content producers to broadcast directly onto televisions through the web.
The Xbox, with over 40 million members, already provides HBO, Netflix, and an array of other premium content directly to televisions via the Internet. Sony\u2019s Playstation does the same! AppleTV, Roku, and Boxee all offer boxtop solutions that directly link TV sets to the web. And major TV companies such as Samsung, LG, and Sony have partnered with Google to directly integrate GoogleTV software into their sets.
Apple is even slated to announce a new Apple Television set (iTV) that will fully integrate Apple\u2019s mobile application store withtelevision sets. Think of the TV being one giant iPad, with the functionality to watch live TV, play games, and stream movies. It only took Apple two years to destroy the tablet market, think about what they can do in the TV world.
The revolution is here folks, and it\u2019s just a matter of time before content producers jump on the bandwagon. Why would big players like Dick Wolf and Jerry Bruckhiemer jump through hoops for major networks when they can distribute their content directly to consumers. They have millions of dollars in the bank to finance their projects, and by selling their content directly to viewers online, they can reel in 100% of the advertising profits. Did you hear me? 100% ad revenue!!!
Simon Fuller, creator/producer of American Idol, has the ability to make $7.1 million dollars an episode in ad revenue if he were to directly sell is show to consumers. American Idol broadcasts roughly 40 episodes a season, which would net him nearly $284 million a season. Now assuming that production costs are maybe $5 million a year, he would stand to be making almost $280 million dollars a season.
But in reality, Simon Fuller sold his entertainment company, Entertainment 19, for a total of $200 million dollars. His company not only owned American Idol, but all of the international licensing rights for the shows concept, which generates millions around the world. And he sold it for less than the ad revenue he would make in one season if he directly distributed online.
Now these numbers are all estimates, but you get the picture. There\u2019s a lot of money left on the table, and that\u2019s because he has to deal with the network. Take that away, and he\u2019s in complete control.
Steve Jobs realized this years ago, when he stopped distributing core Apple products at other retailers. He didn\u2019t see why he was spending so much time and money developing and producing amazing products while companies like Best Buy and Target made money on the sale. It\u2019s why he developed the Apple Store concept, which has become a staple in the company image. If you create a great product, you should sell it, don\u2019t let others take advantage of you.
TV producers need to think the same way. You make the content, you should distribute it. No more middle men, no more jumping through hoops. This is the time to take control, while the Internet is young and untamed. You can take back what is yours, and eliminate the monopolies that have controlled the industry for decades.
The time is now, so make your move.","title":"The End of the Road for the actual American pastime\u2026TV","updated_at":"2023-09-06T20:32:54Z","url":"http://technovisions.wordpress.com/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"camjw"},"story_text":{"matchLevel":"none","matchedWords":[],"value":"Hey everyone,
I'm Cameron and I'm building Sonata, which is a knowledge graph for internal LLM tools.
My thinking is that internal tools are going to become more and more prevalent and complicated with LLMs as it becomes easier to generate high quality software. A core piece of infrastructure that we'll need for this to work is a high quality knowledge graph i.e. a place where all of the information, preferences, goals, terminology etc lives.
I think the future will be that at every company:\n1. There will be a load of really great internal tools (e.g. Klarna replacing Salesforce and Workday with internal tools)\n2. Your single sign on for these tools will come with this knowledge graph.
Another way to think about this is that current "AI Workers" are always on their first day in the job, but with Sonata they'll start on day 100 and only get better with time.
Would love any feedback but more importantly I'd love to demo anyone the product and get you using it.
Thanks!"},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Sonata - platform for self-improving internal tools"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.sonatahq.com/"}},"_tags":["story","author_camjw","story_41873516","show_hn"],"author":"camjw","created_at":"2024-10-17T20:37:58Z","created_at_i":1729197478,"num_comments":0,"objectID":"41873516","points":1,"story_id":41873516,"story_text":"Hey everyone,
I'm Cameron and I'm building Sonata, which is a knowledge graph for internal LLM tools.
My thinking is that internal tools are going to become more and more prevalent and complicated with LLMs as it becomes easier to generate high quality software. A core piece of infrastructure that we'll need for this to work is a high quality knowledge graph i.e. a place where all of the information, preferences, goals, terminology etc lives.
I think the future will be that at every company:\n1. There will be a load of really great internal tools (e.g. Klarna replacing Salesforce and Workday with internal tools)\n2. Your single sign on for these tools will come with this knowledge graph.
Another way to think about this is that current "AI Workers" are always on their first day in the job, but with Sonata they'll start on day 100 and only get better with time.
Would love any feedback but more importantly I'd love to demo anyone the product and get you using it.
Thanks!","title":"Show HN: Sonata - platform for self-improving internal tools","updated_at":"2024-11-08T15:58:09Z","url":"https://www.sonatahq.com/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"digitcatphd"},"story_text":{"matchLevel":"none","matchedWords":[],"value":"The ongoing debate about the future of LLMs often leans heavily towards open-source models. However, it's crucial to pause and consider the merits of private foundation models like GPT, especially in certain scenarios for the foreseeable future.
1. Pricing Dynamics: Concerns about predatory pricing by big players are prevalent. Yet, it's unlikely these companies would adopt such strategies, as they risk driving users away. History shows us that as infrastructure becomes more cost-effective and competition increases, prices tend to decrease, not rise. In specific cases like GitHub CoPilot, high pricing could be a barrier. Still, the real question is whether cutting costs is a viable strategy for maintaining competitiveness.
2. Competitive Edge through 'Secret Sauce': Claiming a unique advantage through a proprietary model essentially positions a company against giants like Microsoft and Google. This advantage, often based on exclusive data and tailored adjustments, might not be sustainable. Moreover, any minor enhancements might be imperceptible to users, leading to a situation akin to building a slightly better version of an existing solution.
3. Focus and Resource Allocation: Dedicating extensive resources to train or fine-tune your open source model might detract from developing other innovative features. In the vast landscape of software and AI, unique functionalities surrounding your product can be more valuable. Without these, a product risks being reduced to a mere interface for an enhanced open-source model.
4. The Quality of Paid Models: While open-source models benefit from a vast community of contributors, the quality of output from a select, highly compensated team at a private firm could be superior. If the top 0.01% of contributors from the open-source community were offered significant compensation by leading tech firms, it's likely they would accept. This suggests that while the open-source community is larger, private firms may have a smaller, yet more distinguished team.
5. Evolving Benchmarks and Release Cycles: There's a common belief that open-source models are quickly catching up to industry benchmarks. However, this perspective doesn't fully account for the differing update cycles between open-source and private foundation models. Open-source models often undergo continual, incremental updates, leading to more frequent but smaller advancements. In contrast, private foundation models like GPT are typically updated in more significant leaps, scheduled at less frequent intervals. This means that when benchmarks are conducted, they often compare the latest open-source models against slightly older versions of private models. Consequently, the benchmarks may not accurately reflect the current capabilities of private models, which could be undergoing substantial advancements behind the scenes, unknown to the public. This discrepancy in update cycles and the nature of advancements can skew perceptions of the actual progress and capabilities of private foundation models compared to their open-source counterparts.
The intention here isn't to undermine open-source models but to present counterarguments to the predominant trend of gravitating towards them. It's important to maintain a balanced perspective and recognize the potential of private foundation models in the evolving landscape of LLMs."},"title":{"matchLevel":"none","matchedWords":[],"value":"Reevaluating the Role of Private Foundation Models in the LLM Race"}},"_tags":["story","author_digitcatphd","story_38261956","ask_hn"],"author":"digitcatphd","created_at":"2023-11-14T11:28:59Z","created_at_i":1699961339,"num_comments":0,"objectID":"38261956","points":1,"story_id":38261956,"story_text":"The ongoing debate about the future of LLMs often leans heavily towards open-source models. However, it's crucial to pause and consider the merits of private foundation models like GPT, especially in certain scenarios for the foreseeable future.
1. Pricing Dynamics: Concerns about predatory pricing by big players are prevalent. Yet, it's unlikely these companies would adopt such strategies, as they risk driving users away. History shows us that as infrastructure becomes more cost-effective and competition increases, prices tend to decrease, not rise. In specific cases like GitHub CoPilot, high pricing could be a barrier. Still, the real question is whether cutting costs is a viable strategy for maintaining competitiveness.
2. Competitive Edge through 'Secret Sauce': Claiming a unique advantage through a proprietary model essentially positions a company against giants like Microsoft and Google. This advantage, often based on exclusive data and tailored adjustments, might not be sustainable. Moreover, any minor enhancements might be imperceptible to users, leading to a situation akin to building a slightly better version of an existing solution.
3. Focus and Resource Allocation: Dedicating extensive resources to train or fine-tune your open source model might detract from developing other innovative features. In the vast landscape of software and AI, unique functionalities surrounding your product can be more valuable. Without these, a product risks being reduced to a mere interface for an enhanced open-source model.
4. The Quality of Paid Models: While open-source models benefit from a vast community of contributors, the quality of output from a select, highly compensated team at a private firm could be superior. If the top 0.01% of contributors from the open-source community were offered significant compensation by leading tech firms, it's likely they would accept. This suggests that while the open-source community is larger, private firms may have a smaller, yet more distinguished team.
5. Evolving Benchmarks and Release Cycles: There's a common belief that open-source models are quickly catching up to industry benchmarks. However, this perspective doesn't fully account for the differing update cycles between open-source and private foundation models. Open-source models often undergo continual, incremental updates, leading to more frequent but smaller advancements. In contrast, private foundation models like GPT are typically updated in more significant leaps, scheduled at less frequent intervals. This means that when benchmarks are conducted, they often compare the latest open-source models against slightly older versions of private models. Consequently, the benchmarks may not accurately reflect the current capabilities of private models, which could be undergoing substantial advancements behind the scenes, unknown to the public. This discrepancy in update cycles and the nature of advancements can skew perceptions of the actual progress and capabilities of private foundation models compared to their open-source counterparts.
The intention here isn't to undermine open-source models but to present counterarguments to the predominant trend of gravitating towards them. It's important to maintain a balanced perspective and recognize the potential of private foundation models in the evolving landscape of LLMs.","title":"Reevaluating the Role of Private Foundation Models in the LLM Race","updated_at":"2024-09-20T15:39:37Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"czsun"},"story_text":{"matchLevel":"none","matchedWords":[],"value":"This is the third post in my ongoing series critically examining the Fugue paper's assertions about Operational Transformation (OT). In the previous two posts, I conducted a thorough analysis of the Fugue paper's arguments, meticulously highlighting the flaws in its reasoning and exposing inaccuracies in its depiction of OT algorithms, particularly the adOPTed algorithm and Jupiter-OT.
To recap, my first post titled "What's Wrong with 'The Art of the Fugue' Paper about OT (adOPTed)?" (https://news.ycombinator.com/item?id=36208585) presented a comprehensive analysis showcasing the consistent and non-interleaving outcomes delivered by the adOPTed algorithm, thereby refuting the alleged "char-interleaving" problem. Moreover, I revealed a fundamental flaw in the Fugue paper's portrayal of the adOPTed algorithm\u2014it mistakenly presented a flawed dOPT-like algorithm instead of the authentic adOPTed algorithm, disregarding the resolution of the well-known dOPT-puzzle. It is disheartening to witness the perpetuation of the dOPT-puzzle within the pages of the Fugue paper, despite its long-standing resolution.
In my second post titled "Unveiling Issues with 'The Art of the Fugue' Paper Regarding Jupiter-OT" (https://news.ycombinator.com/item?id=36415068), I provided a comprehensive explanation of why Jupiter-OT consistently produces non-interleaving outcomes, irrespective of whether it is utilized with string-wise or char-wise transformation functions. This effectively debunked the Fugue paper's baseless claims about Jupiter-OT's "char-interleaving" problem. Additionally, I questioned the relevance and value of discussing concepts like "multi-user-backward-relay-interleaving," urging to direct collective efforts towards addressing genuine co-editing challenges for the advancement of the field.
In this third post, I focus on debunking the unfounded assertions made in the Fugue paper regarding the GOT algorithm. Since GOT supports string-wise co-editing, like Jupiter-OT, and can be combined with various transformation functions, it is straightforward to refute the alleged "char-interleaving" problem in GOT using the same reasoning and illustrations from my second post on Jupiter-OT. Therefore, this post aims to address broader issues, dispel misconceptions, and unveil the truth about the GOT algorithm and OT as a whole.
1.Basic Facts and Features of the GOT algorithm
The GOT (Generic Operation Transformation) work was mainly motivated to solve the classic dOPT puzzle. The GOT algorithm was initially designed and published in [1], without reference to any concrete transformation functions. Later, the combination of the GOT algorithm with a set of independently designed string-wise transformation functions was published in [2].
The GOT algorithm possesses the following main features:
a. Functioning as a distributed OT control algorithm, without relying on a central transformation server.
b.Introducing the notion of operation context and context-based transformation conditions for OT correctness.
c.Solving the dOPT puzzle by ensuring the context-equivalence condition.
d.Achieving convergence without requiring the supporting transformation functions to meet CP1 and CP2 transformation properties.
e.Incorporating a state-vector-based garbage collection scheme to remove operations from the history buffer that are no longer necessary for future transformation.
Similar to Jupiter-OT and the adOPTed algorithm, the GOT algorithm satisfies the mandatory context-based conditions required for all OT control algorithms (see Q&A 3.15-3.18 in OTFAQ [4]); and it can be combined with any suitable transformation functions (not limited to those published in [2]) to create a complete OT solution.
Differing from Jupiter-OT and the adOPTed algorithm, the GOT algorithm employs a pair of Inclusion and Exclusion transformation functions, which are obligated to meet a reversibility transformation property. This reversibility requirement increases the complexity of transformation functions and has been eliminated in subsequent OT control algorithms such as NICE, TIBOT, COT, and POT, which exclusively utilize Inclusion transformation functions.
One side-product of the GOT work is the identification of the False-Tie (FT) puzzle in text co-editing, which has influenced subsequent development in OT and the first CRDT (WOOT) in co-editing. The FT puzzle and CP2-voilation issue have been solved in numerous ways under the OT framework. Readers interested in learning more about FT and its solutions can refer to the following Q&A entries in the OTFAQ [4]:
\u20223.24. What is the False-Tie (FT) puzzle?
\u20223.25. Under what circumstances is an FT-solution needed or not needed?
\u20223.26. How to achieve consistency without solving the FT puzzle?
2. Text-Interleaving is Prohibited in String-Wise Transformation Functions
In the Fugue paper, it was claimed that the "interleaving" problem "has gone unnoticed for decades." However, as I highlighted in my first post, the issue of char-interleaving in some CRDT algorithms (e.g., Logoot) had already been reported as early as 2018. Furthermore, it is important to note that the matter of avoiding concurrent insertion interleaving had been explicitly addressed back in 1998 when designing string-wise transformation functions.
Section 9.1.3 "Criteria for Verifying Intention-Preserved Effects" of [2] (pp. 85-86) provides a precise specification for achieving intention-preserved effects during concurrent string-wise insert and delete operations. This specification served as a guiding principle for the design of string-wise transformation functions, which aim to achieve desired combined effects while explicitly preventing the "interleaving" of concurrent insertions. The following excerpt from [2] highlights this point:
"When the above criteria are satisfied, the execution effects of independent Insert/Delete operations will not interfere with each other in the following sense: an Insert operation may never insert a string into the middle of another string inserted by an independent operation, and a Delete operation may never delete characters inserted by independent operations."\n3.Text-Interleaving is Irrelevant to OT Control Algorithms
Text-interleaving is a special concern in text co-editing. It is a common misconception in some co-editing articles to attribute text co-editing issues to generic OT control algorithms.
In the Fugue paper, Jupiter-OT, adOPTed, and GOT are implicated as the cause of text-interleaving problems. However, even if those illustrations used to support such assertions were valid (which, as demonstrated in my previous posts, they are not), assigning the responsibility of text-editing specific issues to OT control algorithms is misguided and highly misleading. The correctness of an OT control algorithm is determined by its adherence to essential context-based transformation conditions. These conditions are entirely unrelated to text-editing and, consequently, text-interleaving.
This further underscores the need for a better understanding of the principles that govern OT control algorithms and their evaluation criteria. Readers interested in learning more about OT correctness can refer to the following Q&A entries in the OTFAQ [4]:
\u20223.15. What are the OT algorithm correctness requirements?
\u20223.16. Which OT components are responsible for meeting specific algorithm correctness requirements?
\u20223.18. Under what conditions is an OT system algorithmically correct?
4.How to Create Correct OT Solutions by Combining Existing Control Algorithms and Transformation Functions?
A well-established approach to constructing a comprehensive OT solution involves the separation of high-level OT control algorithms from low-level transformation functions, with the specification of their interrelationships through transformation properties and conditions.
One significant advantage of this modular OT system structure is the ability to design and validate control algorithms and transformation functions independently, enabling their flexible combination to create new OT solutions tailored to specific applications, as long as they adhere to the required transformation conditions and properties. The separation and flexible combination of control algorithms and transformation functions have greatly contributed to the continuous advancement of OT and its diverse real-world applications.
Last decade has witnessed significant expansion of OT into new co-editing domains through the invention of novel transformation functions for various data types, such as QuillJS OT functions for rich-text co-editing (https://github.com/ottypes/rich-text), JSON OT functions (https://github.com/ottypes/json0), just to mention a few. Many of these novel transformation functions have been developed by open-source contributors and industry practitioners.
On the other hand, numerous OT control algorithms have been designed and most of them are invented by academic researchers [4]. Some control algorithms, like Jupiter-OT, NICE and Google OT, are Sever-based OT (SOT) algorithms that rely on a central transformation server. However, most other OT control algorithms, including adOPTed, GOT, GOTO, COT, SOCT, TIBOT, and POT, are Distributed OT (DOT) algorithms that do not require a transformation server and allow co-editing clients to connect with each other in flexible communication topologies.
With the availability of a range of OT control algorithms and open-source transformation functions, there are ample opportunities to create comprehensive OT solutions for specific applications by flexibly combining suitable control algorithms and transformation functions.
However, there is a prevalent misconception within co-editing communities that OT necessitates a central server to function. This widespread illusion can be attributed to a combination of factors, including the fact that the popular OT-based Google Docs utilizes a transformation server, a general lack of awareness and understanding of distributed OT algorithms, and the spread of misinformation. Even among experienced industrial engineers and open-source practitioners who have developed practical OT-based co-editing products or designed advanced transformation functions, there was a lack of awareness or limited knowledge about the fact that OT can function perfectly without relying on a central server. This lack of awareness and understanding, combined with the prevailing misconception, led them to mistakenly perceive that their OT systems or functions were confined to operating with a central transformation server like Google Docs.
In fact, OT control algorithms (whether SOT or DOT) and transformation functions (for any data types and applications) are independent components. The publicly available transformation functions developed by practitioners have been commonly integrated with different OT control algorithms (SOT or DOT) in various practical co-editing applications. It is worth noting that most co-editing systems adopt a client-server architecture for valid reasons [3]. If necessary, a server-based OT co-editing system can be transformed into a server-less OT-based co-editing system by adopting a distributed OT control algorithm. This conversion does not require modifying the existing transformation functions for the target application, nor does it necessitate the creation of a new OT control algorithm, as there are numerous existing options readily available.
The notion that OT is unsuitable for peer-to-peer co-editing is a false proposition. For further discussion, refer to Section 4 "Myths and Facts about Peer-to-Peer Co-Editing" in [3].
5. How to Avoid Creating Incorrect OT Solutions in Combining Control Algorithms and Transformation Functions?
While the flexible combination of control algorithms and transformation has been instrumental in creating innovative and effective OT solutions, it is important to acknowledge that this power can, and unfortunately has been, misused to generate incorrect solutions, often employed to substantiate unfounded criticisms of OT. Such misuse may arise from a limited knowledge of OT fundamentals, but its repercussions are far-reaching. It perpetuates distorted views of OT, compromises the integrity of the field, and hinders the overall progress of co-editing.
One example of such misuse can be found in the Fugue paper, which I discussed in detail in my first post of this series. The paper attempted to demonstrate the presence of char-interleaving in the adOPTed algorithm by combining it with the Tombstone Transformation Function (TTF). Unfortunately, the adOPTed algorithm was inaccurately portrayed to function similarly to the flawed dOPT algorithm. This combination of TTF with a dOPT-like algorithm resulted in an erroneous solution that generated inconsistent and interleaving outcomes. These outcomes were then used to support the assertion of an interleaving issue in the adOPTed algorithm and TTF.
In fact, TTF has no connection to char-interleaving either. However, other misconceptions surrounding TTF do exist. In some articles and talks, TTF was portrayed as a correct OT solution, while simultaneously labelling OT control algorithms (such as adOPTed) as incorrect in comparison. However, this comparison is fundamentally flawed because TTF merely comprises a set of transformation functions that must be combined with a suitable OT control algorithm to form a complete solution. Even then, TTF alone does not ensure the correctness of the resulting solution. The Fugue paper serves as a prime example of this, where the combination of TTF with a dOPT-like control algorithm yielded a flawed solution.
Another noteworthy case from the Fugue paper involves the combination of the Jupiter-OT control algorithm with a fabricated char-wise transformation function. This combination was used to justify the alleged issue of char-interleaving within the original Jupiter-OT solution.
In contrast, my second post in this series presented an alternative approach by combining the Jupiter-OT control algorithm with string-wise transformation functions, resulting in consistent and non-interleaving outcomes. Additionally, I presented another new OT solution by integrating the Jupiter-OT control algorithm with a different char-wise transformation function. This solution successfully generated consistent and non-interleaving results for concurrent char-wise insertions.
The moral of the story is clear: the power of combining OT control algorithms and transformation functions in the field of co-editing is immense, but it should be used constructively and responsibly. To harness this power effectively, it is crucial to have a better and more comprehensive understanding of the fundamentals of OT. By doing so, we can avoid potential pitfalls and accelerate the development of correct, valuable, and robust co-editing solutions that drive meaningful progress in the field.
References:
[1] C. Sun, X. Jia, Y. Zhang and Y. Yang: \u201cA Generic Operation Transformation Scheme for Consistency Maintenance in Real-time Cooperative Editing Systems,\u201d Proc. of ACM Conf. on Supporting Group Work, pp. 425 \u2013 434, Nov. 16 \u2013 19, 1997.
[2] C. Sun, X. Jia, Y. Zhang, Y. Yang and D. Chen: "Achieving convergence, causality-preservation, and intention-preservation in real-time cooperative editing systems," ACM Transactions on Computer-Human Interaction, Vol. 5, No. 1, pp.63 \u2013 108, Mar., 1998.
[3] D. Sun, C. Sun, Agustina, W. Cai. Real differences between OT and CRDT in building co-editing systems and real-world applications. https://arxiv.org/abs/1905.01517, May 2, 2019.
[4] C. Sun, "OTFAQ: Operational Transformation Frequently Asked Questions and Answers," https://www3.ntu.edu.sg/scse/staff/czsun/projects/otfaq/
Readers are encouraged to contact the author of this post for copies of any articles referred in this series."},"title":{"matchLevel":"none","matchedWords":[],"value":"Dispelling Misconceptions and Unveiling the Truth about GOT and OT in General"}},"_tags":["story","author_czsun","story_36643393","ask_hn"],"author":"czsun","created_at":"2023-07-08T11:27:45Z","created_at_i":1688815665,"num_comments":0,"objectID":"36643393","points":1,"story_id":36643393,"story_text":"This is the third post in my ongoing series critically examining the Fugue paper's assertions about Operational Transformation (OT). In the previous two posts, I conducted a thorough analysis of the Fugue paper's arguments, meticulously highlighting the flaws in its reasoning and exposing inaccuracies in its depiction of OT algorithms, particularly the adOPTed algorithm and Jupiter-OT.
To recap, my first post titled "What's Wrong with 'The Art of the Fugue' Paper about OT (adOPTed)?" (https://news.ycombinator.com/item?id=36208585) presented a comprehensive analysis showcasing the consistent and non-interleaving outcomes delivered by the adOPTed algorithm, thereby refuting the alleged "char-interleaving" problem. Moreover, I revealed a fundamental flaw in the Fugue paper's portrayal of the adOPTed algorithm\u2014it mistakenly presented a flawed dOPT-like algorithm instead of the authentic adOPTed algorithm, disregarding the resolution of the well-known dOPT-puzzle. It is disheartening to witness the perpetuation of the dOPT-puzzle within the pages of the Fugue paper, despite its long-standing resolution.
In my second post titled "Unveiling Issues with 'The Art of the Fugue' Paper Regarding Jupiter-OT" (https://news.ycombinator.com/item?id=36415068), I provided a comprehensive explanation of why Jupiter-OT consistently produces non-interleaving outcomes, irrespective of whether it is utilized with string-wise or char-wise transformation functions. This effectively debunked the Fugue paper's baseless claims about Jupiter-OT's "char-interleaving" problem. Additionally, I questioned the relevance and value of discussing concepts like "multi-user-backward-relay-interleaving," urging to direct collective efforts towards addressing genuine co-editing challenges for the advancement of the field.
In this third post, I focus on debunking the unfounded assertions made in the Fugue paper regarding the GOT algorithm. Since GOT supports string-wise co-editing, like Jupiter-OT, and can be combined with various transformation functions, it is straightforward to refute the alleged "char-interleaving" problem in GOT using the same reasoning and illustrations from my second post on Jupiter-OT. Therefore, this post aims to address broader issues, dispel misconceptions, and unveil the truth about the GOT algorithm and OT as a whole.
1.Basic Facts and Features of the GOT algorithm
The GOT (Generic Operation Transformation) work was mainly motivated to solve the classic dOPT puzzle. The GOT algorithm was initially designed and published in [1], without reference to any concrete transformation functions. Later, the combination of the GOT algorithm with a set of independently designed string-wise transformation functions was published in [2].
The GOT algorithm possesses the following main features:
a. Functioning as a distributed OT control algorithm, without relying on a central transformation server.
b.Introducing the notion of operation context and context-based transformation conditions for OT correctness.
c.Solving the dOPT puzzle by ensuring the context-equivalence condition.
d.Achieving convergence without requiring the supporting transformation functions to meet CP1 and CP2 transformation properties.
e.Incorporating a state-vector-based garbage collection scheme to remove operations from the history buffer that are no longer necessary for future transformation.
Similar to Jupiter-OT and the adOPTed algorithm, the GOT algorithm satisfies the mandatory context-based conditions required for all OT control algorithms (see Q&A 3.15-3.18 in OTFAQ [4]); and it can be combined with any suitable transformation functions (not limited to those published in [2]) to create a complete OT solution.
Differing from Jupiter-OT and the adOPTed algorithm, the GOT algorithm employs a pair of Inclusion and Exclusion transformation functions, which are obligated to meet a reversibility transformation property. This reversibility requirement increases the complexity of transformation functions and has been eliminated in subsequent OT control algorithms such as NICE, TIBOT, COT, and POT, which exclusively utilize Inclusion transformation functions.
One side-product of the GOT work is the identification of the False-Tie (FT) puzzle in text co-editing, which has influenced subsequent development in OT and the first CRDT (WOOT) in co-editing. The FT puzzle and CP2-voilation issue have been solved in numerous ways under the OT framework. Readers interested in learning more about FT and its solutions can refer to the following Q&A entries in the OTFAQ [4]:
\u20223.24. What is the False-Tie (FT) puzzle?
\u20223.25. Under what circumstances is an FT-solution needed or not needed?
\u20223.26. How to achieve consistency without solving the FT puzzle?
2. Text-Interleaving is Prohibited in String-Wise Transformation Functions
In the Fugue paper, it was claimed that the "interleaving" problem "has gone unnoticed for decades." However, as I highlighted in my first post, the issue of char-interleaving in some CRDT algorithms (e.g., Logoot) had already been reported as early as 2018. Furthermore, it is important to note that the matter of avoiding concurrent insertion interleaving had been explicitly addressed back in 1998 when designing string-wise transformation functions.
Section 9.1.3 "Criteria for Verifying Intention-Preserved Effects" of [2] (pp. 85-86) provides a precise specification for achieving intention-preserved effects during concurrent string-wise insert and delete operations. This specification served as a guiding principle for the design of string-wise transformation functions, which aim to achieve desired combined effects while explicitly preventing the "interleaving" of concurrent insertions. The following excerpt from [2] highlights this point:
"When the above criteria are satisfied, the execution effects of independent Insert/Delete operations will not interfere with each other in the following sense: an Insert operation may never insert a string into the middle of another string inserted by an independent operation, and a Delete operation may never delete characters inserted by independent operations."\n3.Text-Interleaving is Irrelevant to OT Control Algorithms
Text-interleaving is a special concern in text co-editing. It is a common misconception in some co-editing articles to attribute text co-editing issues to generic OT control algorithms.
In the Fugue paper, Jupiter-OT, adOPTed, and GOT are implicated as the cause of text-interleaving problems. However, even if those illustrations used to support such assertions were valid (which, as demonstrated in my previous posts, they are not), assigning the responsibility of text-editing specific issues to OT control algorithms is misguided and highly misleading. The correctness of an OT control algorithm is determined by its adherence to essential context-based transformation conditions. These conditions are entirely unrelated to text-editing and, consequently, text-interleaving.
This further underscores the need for a better understanding of the principles that govern OT control algorithms and their evaluation criteria. Readers interested in learning more about OT correctness can refer to the following Q&A entries in the OTFAQ [4]:
\u20223.15. What are the OT algorithm correctness requirements?
\u20223.16. Which OT components are responsible for meeting specific algorithm correctness requirements?
\u20223.18. Under what conditions is an OT system algorithmically correct?
4.How to Create Correct OT Solutions by Combining Existing Control Algorithms and Transformation Functions?
A well-established approach to constructing a comprehensive OT solution involves the separation of high-level OT control algorithms from low-level transformation functions, with the specification of their interrelationships through transformation properties and conditions.
One significant advantage of this modular OT system structure is the ability to design and validate control algorithms and transformation functions independently, enabling their flexible combination to create new OT solutions tailored to specific applications, as long as they adhere to the required transformation conditions and properties. The separation and flexible combination of control algorithms and transformation functions have greatly contributed to the continuous advancement of OT and its diverse real-world applications.
Last decade has witnessed significant expansion of OT into new co-editing domains through the invention of novel transformation functions for various data types, such as QuillJS OT functions for rich-text co-editing (https://github.com/ottypes/rich-text), JSON OT functions (https://github.com/ottypes/json0), just to mention a few. Many of these novel transformation functions have been developed by open-source contributors and industry practitioners.
On the other hand, numerous OT control algorithms have been designed and most of them are invented by academic researchers [4]. Some control algorithms, like Jupiter-OT, NICE and Google OT, are Sever-based OT (SOT) algorithms that rely on a central transformation server. However, most other OT control algorithms, including adOPTed, GOT, GOTO, COT, SOCT, TIBOT, and POT, are Distributed OT (DOT) algorithms that do not require a transformation server and allow co-editing clients to connect with each other in flexible communication topologies.
With the availability of a range of OT control algorithms and open-source transformation functions, there are ample opportunities to create comprehensive OT solutions for specific applications by flexibly combining suitable control algorithms and transformation functions.
However, there is a prevalent misconception within co-editing communities that OT necessitates a central server to function. This widespread illusion can be attributed to a combination of factors, including the fact that the popular OT-based Google Docs utilizes a transformation server, a general lack of awareness and understanding of distributed OT algorithms, and the spread of misinformation. Even among experienced industrial engineers and open-source practitioners who have developed practical OT-based co-editing products or designed advanced transformation functions, there was a lack of awareness or limited knowledge about the fact that OT can function perfectly without relying on a central server. This lack of awareness and understanding, combined with the prevailing misconception, led them to mistakenly perceive that their OT systems or functions were confined to operating with a central transformation server like Google Docs.
In fact, OT control algorithms (whether SOT or DOT) and transformation functions (for any data types and applications) are independent components. The publicly available transformation functions developed by practitioners have been commonly integrated with different OT control algorithms (SOT or DOT) in various practical co-editing applications. It is worth noting that most co-editing systems adopt a client-server architecture for valid reasons [3]. If necessary, a server-based OT co-editing system can be transformed into a server-less OT-based co-editing system by adopting a distributed OT control algorithm. This conversion does not require modifying the existing transformation functions for the target application, nor does it necessitate the creation of a new OT control algorithm, as there are numerous existing options readily available.
The notion that OT is unsuitable for peer-to-peer co-editing is a false proposition. For further discussion, refer to Section 4 "Myths and Facts about Peer-to-Peer Co-Editing" in [3].
5. How to Avoid Creating Incorrect OT Solutions in Combining Control Algorithms and Transformation Functions?
While the flexible combination of control algorithms and transformation has been instrumental in creating innovative and effective OT solutions, it is important to acknowledge that this power can, and unfortunately has been, misused to generate incorrect solutions, often employed to substantiate unfounded criticisms of OT. Such misuse may arise from a limited knowledge of OT fundamentals, but its repercussions are far-reaching. It perpetuates distorted views of OT, compromises the integrity of the field, and hinders the overall progress of co-editing.
One example of such misuse can be found in the Fugue paper, which I discussed in detail in my first post of this series. The paper attempted to demonstrate the presence of char-interleaving in the adOPTed algorithm by combining it with the Tombstone Transformation Function (TTF). Unfortunately, the adOPTed algorithm was inaccurately portrayed to function similarly to the flawed dOPT algorithm. This combination of TTF with a dOPT-like algorithm resulted in an erroneous solution that generated inconsistent and interleaving outcomes. These outcomes were then used to support the assertion of an interleaving issue in the adOPTed algorithm and TTF.
In fact, TTF has no connection to char-interleaving either. However, other misconceptions surrounding TTF do exist. In some articles and talks, TTF was portrayed as a correct OT solution, while simultaneously labelling OT control algorithms (such as adOPTed) as incorrect in comparison. However, this comparison is fundamentally flawed because TTF merely comprises a set of transformation functions that must be combined with a suitable OT control algorithm to form a complete solution. Even then, TTF alone does not ensure the correctness of the resulting solution. The Fugue paper serves as a prime example of this, where the combination of TTF with a dOPT-like control algorithm yielded a flawed solution.
Another noteworthy case from the Fugue paper involves the combination of the Jupiter-OT control algorithm with a fabricated char-wise transformation function. This combination was used to justify the alleged issue of char-interleaving within the original Jupiter-OT solution.
In contrast, my second post in this series presented an alternative approach by combining the Jupiter-OT control algorithm with string-wise transformation functions, resulting in consistent and non-interleaving outcomes. Additionally, I presented another new OT solution by integrating the Jupiter-OT control algorithm with a different char-wise transformation function. This solution successfully generated consistent and non-interleaving results for concurrent char-wise insertions.
The moral of the story is clear: the power of combining OT control algorithms and transformation functions in the field of co-editing is immense, but it should be used constructively and responsibly. To harness this power effectively, it is crucial to have a better and more comprehensive understanding of the fundamentals of OT. By doing so, we can avoid potential pitfalls and accelerate the development of correct, valuable, and robust co-editing solutions that drive meaningful progress in the field.
References:
[1] C. Sun, X. Jia, Y. Zhang and Y. Yang: \u201cA Generic Operation Transformation Scheme for Consistency Maintenance in Real-time Cooperative Editing Systems,\u201d Proc. of ACM Conf. on Supporting Group Work, pp. 425 \u2013 434, Nov. 16 \u2013 19, 1997.
[2] C. Sun, X. Jia, Y. Zhang, Y. Yang and D. Chen: "Achieving convergence, causality-preservation, and intention-preservation in real-time cooperative editing systems," ACM Transactions on Computer-Human Interaction, Vol. 5, No. 1, pp.63 \u2013 108, Mar., 1998.
[3] D. Sun, C. Sun, Agustina, W. Cai. Real differences between OT and CRDT in building co-editing systems and real-world applications. https://arxiv.org/abs/1905.01517, May 2, 2019.
[4] C. Sun, "OTFAQ: Operational Transformation Frequently Asked Questions and Answers," https://www3.ntu.edu.sg/scse/staff/czsun/projects/otfaq/
Readers are encouraged to contact the author of this post for copies of any articles referred in this series.","title":"Dispelling Misconceptions and Unveiling the Truth about GOT and OT in General","updated_at":"2024-09-20T14:35:08Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"techknowlogick"},"title":{"matchLevel":"none","matchedWords":[],"value":"Anubis: Proof-of-work proxy to prevent AI crawlers"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://anubis.techaro.lol/"}},"_tags":["story","author_techknowlogick","story_43427679"],"author":"techknowlogick","children":[43427847,43427881,43427917,43427932,43428020,43428026,43428213,43428214,43428232,43428546,43428692,43428737,43428738,43428766,43429060,43429160,43429538,43430322,43430347,43430396,43430775,43430947,43436818],"created_at":"2025-03-20T19:16:22Z","created_at_i":1742498182,"num_comments":58,"objectID":"43427679","points":100,"story_id":43427679,"title":"Anubis: Proof-of-work proxy to prevent AI crawlers","updated_at":"2026-05-25T19:20:37Z","url":"https://anubis.techaro.lol/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"BaptisteGreve"},"title":{"matchLevel":"none","matchedWords":[],"value":"Google is working on a kill switch to prevent an AI uprising"},"url":{"matchLevel":"none","matchedWords":[],"value":"http://www.engadget.com/2016/06/03/google-ai-killswitch/"}},"_tags":["story","author_BaptisteGreve","story_11835624"],"author":"BaptisteGreve","children":[11835805,11835889,11835894,11835946,11835950,11835955,11836033,11836127,11836142],"created_at":"2016-06-04T09:13:12Z","created_at_i":1465031592,"num_comments":42,"objectID":"11835624","points":27,"story_id":11835624,"title":"Google is working on a kill switch to prevent an AI uprising","updated_at":"2024-09-19T23:17:32Z","url":"http://www.engadget.com/2016/06/03/google-ai-killswitch/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"benkan"},"title":{"matchLevel":"none","matchedWords":[],"value":"Getty Images CEO: Respecting fair use rules won't prevent AI from curing cancer"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://fortune.com/2024/12/02/getty-images-ceo-respecting-fair-use-rules-wont-prevent-ai-from-curing-cancer-tech-law/"}},"_tags":["story","author_benkan","story_42299593"],"author":"benkan","children":[42300010,42301225,42301569,42301665,42301844],"created_at":"2024-12-02T19:41:08Z","created_at_i":1733168468,"num_comments":17,"objectID":"42299593","points":24,"story_id":42299593,"title":"Getty Images CEO: Respecting fair use rules won't prevent AI from curing cancer","updated_at":"2025-08-14T22:27:38Z","url":"https://fortune.com/2024/12/02/getty-images-ceo-respecting-fair-use-rules-wont-prevent-ai-from-curing-cancer-tech-law/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"atothei"},"title":{"matchLevel":"none","matchedWords":[],"value":"63 Percent of Americans want regulation to actively prevent superintelligent AI"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.vox.com/future-perfect/2023/9/19/23879648/americans-artificial-general-intelligence-ai-policy-poll"}},"_tags":["story","author_atothei","story_40353310"],"author":"atothei","children":[40354109,40354332,40354396,40354646,40354658,40354949,40354980,40355102,40355578,40355603,40365488],"created_at":"2024-05-14T09:43:12Z","created_at_i":1715679792,"num_comments":21,"objectID":"40353310","points":23,"story_id":40353310,"title":"63 Percent of Americans want regulation to actively prevent superintelligent AI","updated_at":"2024-09-20T17:03:50Z","url":"https://www.vox.com/future-perfect/2023/9/19/23879648/americans-artificial-general-intelligence-ai-policy-poll"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"georgehill"},"title":{"matchLevel":"none","matchedWords":[],"value":"Microsoft limits Bing chat to five replies to prevent AI from behaving oddly"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.theverge.com/2023/2/17/23604906/microsoft-bing-ai-chat-limits-conversations"}},"_tags":["story","author_georgehill","story_34842121"],"author":"georgehill","children":[34842180,34842312,34842535,34842729,34843016,34844047,34844528],"created_at":"2023-02-17T23:43:45Z","created_at_i":1676677425,"num_comments":9,"objectID":"34842121","points":16,"story_id":34842121,"title":"Microsoft limits Bing chat to five replies to prevent AI from behaving oddly","updated_at":"2024-09-20T13:23:51Z","url":"https://www.theverge.com/2023/2/17/23604906/microsoft-bing-ai-chat-limits-conversations"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"wglb"},"title":{"matchLevel":"none","matchedWords":[],"value":"Nav Canada leads global effort to prevent lost airliners"},"url":{"matchLevel":"none","matchedWords":[],"value":"http://ottawacitizen.com/news/national/nav-canada-leads-global-effort-to-prevent-lost-airliners"}},"_tags":["story","author_wglb","story_8353133"],"author":"wglb","children":[8353781],"created_at":"2014-09-22T22:51:10Z","created_at_i":1411426270,"num_comments":3,"objectID":"8353133","points":14,"story_id":8353133,"title":"Nav Canada leads global effort to prevent lost airliners","updated_at":"2024-09-19T21:07:48Z","url":"http://ottawacitizen.com/news/national/nav-canada-leads-global-effort-to-prevent-lost-airliners"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"kristianp"},"title":{"matchLevel":"none","matchedWords":[],"value":"The New York Times Updates Terms of Service to Prevent AI Scraping Its Content"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.adweek.com/media/the-new-york-times-updates-terms-of-service-to-prevent-ai-scraping-its-content/"}},"_tags":["story","author_kristianp","story_37126807"],"author":"kristianp","created_at":"2023-08-14T21:07:27Z","created_at_i":1692047247,"num_comments":0,"objectID":"37126807","points":13,"story_id":37126807,"title":"The New York Times Updates Terms of Service to Prevent AI Scraping Its Content","updated_at":"2024-09-20T14:47:55Z","url":"https://www.adweek.com/media/the-new-york-times-updates-terms-of-service-to-prevent-ai-scraping-its-content/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"stikit"},"title":{"matchLevel":"none","matchedWords":[],"value":"US lawmakers introduce bill to prevent AI-controlled nuclear launches"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.engadget.com/us-lawmakers-introduce-bill-to-prevent-ai-controlled-nuclear-launches-184727260.html"}},"_tags":["story","author_stikit","story_35758562"],"author":"stikit","children":[35758981,35759391,35759667],"created_at":"2023-04-30T00:13:48Z","created_at_i":1682813628,"num_comments":2,"objectID":"35758562","points":8,"story_id":35758562,"title":"US lawmakers introduce bill to prevent AI-controlled nuclear launches","updated_at":"2024-09-20T14:01:21Z","url":"https://www.engadget.com/us-lawmakers-introduce-bill-to-prevent-ai-controlled-nuclear-launches-184727260.html"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"blindriver"},"story_text":{"matchLevel":"none","matchedWords":[],"value":"How can I prevent AI (ChatGPT, etc) from reading and training on the blog posts, reddit posts, or code that I post online like on Github?
Is there a license that I can use that specifically prohibits AI companies from reading and training off my content? Or do I need to put my content behind a login?"},"title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: How do I prevent AI from reading/training off my content?"}},"_tags":["story","author_blindriver","story_44629683","ask_hn"],"author":"blindriver","children":[44629712,44630015,44630247,44630386,44630646,44637424,44642019,44644369,44646066,44652726],"created_at":"2025-07-20T21:55:55Z","created_at_i":1753048555,"num_comments":8,"objectID":"44629683","points":7,"story_id":44629683,"story_text":"How can I prevent AI (ChatGPT, etc) from reading and training on the blog posts, reddit posts, or code that I post online like on Github?
Is there a license that I can use that specifically prohibits AI companies from reading and training off my content? Or do I need to put my content behind a login?","title":"Ask HN: How do I prevent AI from reading/training off my content?","updated_at":"2025-08-01T16:27:37Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"Mohil_Sharma"},"story_text":{"matchLevel":"none","matchedWords":[],"value":"Hi HN,\nI\u2019m a solo developer and built AgentWatch to solve a problem I kept running into while building AI agents: preventing runaway loops and unexpected LLM spend before requests reach the model.\nAgentWatch sits in front of OpenAI, Anthropic, Gemini, Bedrock, Azure OpenAI, Groq, and others to enforce budgets and runtime policies.\nI\u2019d really appreciate your feedback. If you\u2019re building AI agents, does this solve a problem you\u2019ve experienced? I\u2019d also love to hear what you\u2019d improve or challenge."},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: AgentWatch \u2013 Prevent runaway AI agents with runtime budget enforcement"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://agent-watch.dev/"}},"_tags":["story","author_Mohil_Sharma","story_48706317","show_hn"],"author":"Mohil_Sharma","children":[48706484,48706891,48707211,48708480,48710773,48710839,48725399],"created_at":"2026-06-28T11:09:20Z","created_at_i":1782644960,"num_comments":5,"objectID":"48706317","points":7,"story_id":48706317,"story_text":"Hi HN,\nI\u2019m a solo developer and built AgentWatch to solve a problem I kept running into while building AI agents: preventing runaway loops and unexpected LLM spend before requests reach the model.\nAgentWatch sits in front of OpenAI, Anthropic, Gemini, Bedrock, Azure OpenAI, Groq, and others to enforce budgets and runtime policies.\nI\u2019d really appreciate your feedback. If you\u2019re building AI agents, does this solve a problem you\u2019ve experienced? I\u2019d also love to hear what you\u2019d improve or challenge.","title":"Show HN: AgentWatch \u2013 Prevent runaway AI agents with runtime budget enforcement","updated_at":"2026-07-03T20:42:56Z","url":"https://agent-watch.dev/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"wingrammer"},"story_text":{"matchLevel":"none","matchedWords":[],"value":"AI agents increasingly execute real system actions: issuing refunds, modifying databases, deploying infrastructure, calling external APIs.
Because agents retry steps, re-plan tasks, and run asynchronously, the same action can sometimes execute more than once.
In production systems this can cause duplicate payouts, repeated mutations, or inconsistent state.
Kybernis is a reliability layer that sits at the execution boundary of agent systems.
When an agent calls a tool:
1. execution intent is captured\n2. the action is recorded in an execution ledger\n3. idempotency guarantees are attached\n4. the mutation commits exactly once
Retries become safe.
Kybernis is framework-neutral and works with agent frameworks like LangGraph, AutoGen, CrewAI, or custom systems.
I built this after repeatedly seeing reliability failures when AI agents interacted with production APIs.
Would love feedback from anyone building agent systems."},"title":{"matchLevel":"none","matchedWords":[],"value":"Show HN: Kybernis \u2013 Prevent AI agents from executing the same action twice"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://kybernis.io"}},"_tags":["story","author_wingrammer","story_47267024","show_hn"],"author":"wingrammer","children":[47267063,47268331,47390812],"created_at":"2026-03-05T20:43:38Z","created_at_i":1772743418,"num_comments":3,"objectID":"47267024","points":6,"story_id":47267024,"story_text":"AI agents increasingly execute real system actions: issuing refunds, modifying databases, deploying infrastructure, calling external APIs.
Because agents retry steps, re-plan tasks, and run asynchronously, the same action can sometimes execute more than once.
In production systems this can cause duplicate payouts, repeated mutations, or inconsistent state.
Kybernis is a reliability layer that sits at the execution boundary of agent systems.
When an agent calls a tool:
1. execution intent is captured\n2. the action is recorded in an execution ledger\n3. idempotency guarantees are attached\n4. the mutation commits exactly once
Retries become safe.
Kybernis is framework-neutral and works with agent frameworks like LangGraph, AutoGen, CrewAI, or custom systems.
I built this after repeatedly seeing reliability failures when AI agents interacted with production APIs.
Would love feedback from anyone building agent systems.","title":"Show HN: Kybernis \u2013 Prevent AI agents from executing the same action twice","updated_at":"2026-03-17T11:40:55Z","url":"https://kybernis.io"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"thunderbong"},"title":{"matchLevel":"none","matchedWords":[],"value":"Caddy module to block IPs and prevent AIs from training on your website"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/JasonLovesDoggo/caddy-defender"}},"_tags":["story","author_thunderbong","story_42809204"],"author":"thunderbong","children":[42809218,42835430],"created_at":"2025-01-23T23:43:38Z","created_at_i":1737675818,"num_comments":1,"objectID":"42809204","points":6,"story_id":42809204,"title":"Caddy module to block IPs and prevent AIs from training on your website","updated_at":"2025-01-26T23:29:39Z","url":"https://github.com/JasonLovesDoggo/caddy-defender"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"ekianjo"},"story_text":{"matchLevel":"none","matchedWords":[],"value":"Hi, since conversation between human users at HN is what makes the platform attractive at it is, I am wondering what HN is doing, or is going to do, when more sophisticated bots could land to parrot certain narratives or influence the conversation on important topics. Is there any way to detect such behavior?"},"title":{"matchLevel":"none","matchedWords":[],"value":"Ask HN: Does HN do anything to prevent AI-grade bots?"}},"_tags":["story","author_ekianjo","story_32124002","ask_hn"],"author":"ekianjo","children":[32124154,32124298,32124501,32124592,32124622,32124652],"created_at":"2022-07-17T02:46:32Z","created_at_i":1658025992,"num_comments":9,"objectID":"32124002","points":5,"story_id":32124002,"story_text":"Hi, since conversation between human users at HN is what makes the platform attractive at it is, I am wondering what HN is doing, or is going to do, when more sophisticated bots could land to parrot certain narratives or influence the conversation on important topics. Is there any way to detect such behavior?","title":"Ask HN: Does HN do anything to prevent AI-grade bots?","updated_at":"2024-09-20T11:35:25Z"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dougb5"},"title":{"matchLevel":"none","matchedWords":[],"value":"63 percent of US voters want regulation to prevent AI superintelligence"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.vox.com/future-perfect/2023/9/19/23879648/americans-artificial-general-intelligence-ai-policy-poll"}},"_tags":["story","author_dougb5","story_37580281"],"author":"dougb5","children":[37580390,37580672],"created_at":"2023-09-20T04:08:13Z","created_at_i":1695182893,"num_comments":2,"objectID":"37580281","points":5,"story_id":37580281,"title":"63 percent of US voters want regulation to prevent AI superintelligence","updated_at":"2024-09-20T15:15:01Z","url":"https://www.vox.com/future-perfect/2023/9/19/23879648/americans-artificial-general-intelligence-ai-policy-poll"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"belter"},"title":{"matchLevel":"none","matchedWords":[],"value":"Hollywood's writers are striking to prevent an AI-generated creative hellscape"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.gq-magazine.co.uk/article/wga-writers-strike-hollywood-ai-2023"}},"_tags":["story","author_belter","story_35802270"],"author":"belter","children":[35802342,35802815],"created_at":"2023-05-03T13:41:53Z","created_at_i":1683121313,"num_comments":2,"objectID":"35802270","points":4,"story_id":35802270,"title":"Hollywood's writers are striking to prevent an AI-generated creative hellscape","updated_at":"2024-09-20T13:56:49Z","url":"https://www.gq-magazine.co.uk/article/wga-writers-strike-hollywood-ai-2023"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"visha1v"},"title":{"matchLevel":"none","matchedWords":[],"value":"Can AI Prevent Suicides?"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://cacm.acm.org/news/can-ai-prevent-suicides/"}},"_tags":["story","author_visha1v","story_48765825"],"author":"visha1v","children":[48766003],"created_at":"2026-07-02T18:56:00Z","created_at_i":1783018560,"num_comments":1,"objectID":"48765825","points":4,"story_id":48765825,"title":"Can AI Prevent Suicides?","updated_at":"2026-07-02T21:35:51Z","url":"https://cacm.acm.org/news/can-ai-prevent-suicides/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"huzpsb"},"title":{"matchLevel":"none","matchedWords":[],"value":"Anubis Solver: Can Anubis Prevent AI Crawlers?"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://github.com/huzpsb/anubis_solver_pypi"}},"_tags":["story","author_huzpsb","story_45247579"],"author":"huzpsb","children":[45247580,45249014],"created_at":"2025-09-15T08:59:40Z","created_at_i":1757926780,"num_comments":1,"objectID":"45247579","points":4,"story_id":45247579,"title":"Anubis Solver: Can Anubis Prevent AI Crawlers?","updated_at":"2026-03-05T22:41:26Z","url":"https://github.com/huzpsb/anubis_solver_pypi"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"dangoldbj"},"title":{"matchLevel":"none","matchedWords":[],"value":"OpenAI and Anthropic Sign Letter to Prevent AI-Developed Biological Weapons"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://www.wired.com/story/openai-anthropic-letter-ai-biological-weapons/"}},"_tags":["story","author_dangoldbj","story_48395821"],"author":"dangoldbj","created_at":"2026-06-04T08:30:22Z","created_at_i":1780561822,"num_comments":0,"objectID":"48395821","points":4,"story_id":48395821,"title":"OpenAI and Anthropic Sign Letter to Prevent AI-Developed Biological Weapons","updated_at":"2026-06-04T11:12:02Z","url":"https://www.wired.com/story/openai-anthropic-letter-ai-biological-weapons/"},{"_highlightResult":{"author":{"matchLevel":"none","matchedWords":[],"value":"trott"},"title":{"matchLevel":"none","matchedWords":[],"value":"From sci-fi to state law: California's plan to prevent AI catastrophe"},"url":{"matchLevel":"none","matchedWords":[],"value":"https://arstechnica.com/information-technology/2024/07/from-sci-fi-to-state-law-californias-plan-to-prevent-ai-catastrophe/"}},"_tags":["story","author_trott","story_41104938"],"author":"trott","created_at":"2024-07-30T00:17:28Z","created_at_i":1722298648,"num_comments":0,"objectID":"41104938","points":4,"story_id":41104938,"title":"From sci-fi to state law: California's plan to prevent AI catastrophe","updated_at":"2024-09-20T17:29:43Z","url":"https://arstechnica.com/information-technology/2024/07/from-sci-fi-to-state-law-californias-plan-to-prevent-ai-catastrophe/"}],"hitsPerPage":50,"nbHits":732,"nbPages":15,"page":0,"params":"query=Prevalent+AI&tags=story&hitsPerPage=50&advancedSyntax=true&analyticsTags=backend","processingTimeMS":11,"processingTimingsMS":{"_request":{"roundTrip":25},"afterFetch":{"format":{"highlighting":5,"total":6},"merge":{"mergeLoop":{"prepareNextHit":1,"total":1},"total":1},"total":1},"fetch":{"query":5,"scanning":3,"total":9},"total":11},"query":"Prevalent AI","serverTimeMS":17}