0%
0/1 resolved calls right
1 scored · avg 20
single source
7
companies · 44 data points
Google will lose a significant amount of revenue if they integrate powerful AI models to provide accurate answers directly in their main search bar.
What happened: Google has not reported a significant revenue loss after integrating AI models into its search bar; overall revenue and ad income have continued to rise.
Verify at source ↗Wow, Claude's having a great run. OpenAI still doing strong. Grok doing very well. Gemini coming on strong.
But what is perplexity in the face of, wow, Claude's having a great run.
AI coding tools have made it much faster for us to ship things, which is honestly one of the reasons why we built computer, because now even non-engineers are shipping code here by just bringing a slack bot and asking it to fix bugs. Yeah. So the iteration has just been like exponential.
Our philosophy is like speed is our mode. Like, you know, again, one of the things that big companies cannot do is move at the speed we do, serve customers at the speed and quality.
the model council is a feature we built where it will not just give you the answers of each model, but it will tell you exactly where they agree, where they disagree and where the nuances are.
the orchestra conductor that has no one model to the horse in the race can win by providing a very unique value and service to the customer that each of these amazing names that you mentioned cannot.
We're like Switzerland. We don't have to have one horse in the race. If GPT wins, Gemini wins, Claude wins, Lama wins. It doesn't matter to us. Or even open source models can win. No problem.
So one advantage we have that all these companies you mentioned don't have is the multi-model orchestration.
My prediction is that initially start off as a sub-agent. So whatever you need to go, like your tax returns, your personal photos, your emails, your calendar, all that stuff, those local apps, your personal notes, very personal notes. You could make sure that the models that access those tokens will be running on your local hardware if you want to, if you're that privacy conscious. And more complicated stuff that accesses your data that's already on the server side, example, your Google Calendar
Nobody needs to manage API keys. Nobody needs to manage separate billing across like a hundred different services, figure out what you can give access to and not access to. We take care of that. So it's a Steve jobs way of doing it.
So we announced something called personal computer, perplexity personal computer. That's essentially going to take all the trust and reliability and the server side execution of perplexity computer, but synchronize it with your local computer so that you can use it from your phone. And we're going to do this with the Mac mini, where you synchronize your computer with the Mac mini. So that becomes your local server. All the agent orchestration that has to do with your local private data will run
One common theme for us for the last three and a half years is accuracy. Perplexity wants to be the company that's building the most accurate AI. So when you want to give somebody answers, accuracy is very essential for building trust. Because only then the user is going to ask the next set of questions. It turns out it was a great idea to give AI access to the internet to be accurate. So that's the perplexity ask product. It turns out it's a great idea for AI to have full access to a browser so
every other organization doesn't want to publish anymore
who can likely catch up there's one more organization called anthropic sure and like they are the closest number two
open ai was ahead of everybody else there um google has all the data in the world but they didn't pay enough attention to this
if these guys who are building the tricks for making these models even better are in the close organizations then they'll always stay ahead of the open right so then and if these organizations stop publishing these techniques um and these guys sustain these organizations are paid to stay there forever um it's kind of like closing the walls
as for like who's really ahead right now i think it's uh open ai like with gpt forward yeah yeah much far ahead who can likely catch up there's one more organization called anthropic sure and like they are the closest number two
the difference comes to what data it's being trained on um where open ai puts in a lot of effort there compared to other organizations
the work being at open ai was really helpful back then there was no chat gpt so i didn't foresee the future where open ai is so successful but there was gpd 3.5 and was pretty good and like we knew like a lot of things were happening nobody knew that like if you put out these models in the chat ui the world would go crazy
we don't waste time trying to hire people that same woman will be hiring anyway it's very it's you cannot compete they have way more cash way more like and they can give way less percent of the company because they have way bigger valuation so
the reason is they have the best models
we heavily use chat gpt 3.5 and 4 and we also use a little bit of our own lms for many other things
both chat gpt and us were basically being powered by gpt 3.5 uh that was the fundamental breakthrough and then after that gpt4 even better than that ... chat gpt was this entertaining hallucinatory bot that was not not just you know like correct many times but it was also equally wrong many times and its mistakes were also entertaining
they still have a lot of distribution through windows devices um so a lot of people just go to edge and they can start using bing chat um but people have despite that we won against microsoft in the past like google everyone went and searched for chrome as the first search query on like internet explorer to install it they did that despite the friction they added
people have done comparisons between us and bing and um you know we have like much better accuracy in terms of how correctly we cite things a lot of academic research has been done there uh people spend on an average like two minutes more on our site than bing uh so our engagement is much better there our bounce rates are much lower than bing
everything is a transformer uh which is the architecture built by google in 2017
they google went and searched for chrome as the first search query on like internet explorer to install it we all did that despite the friction they added
they cannot afford to use really powerful models so like the search traffic that they have and if they actually want to really get it right but uh in the actual search bar itself outside of bard they're
google's also rolling out this thing called magi or they call it search generative experience to the public but uh the wall street journal called it magi um so they're they're trying to do something pretty similar to us and so far the experience at least from what i've seen and myself used that and other people who use that is it's not very different from the way they used to extract text from the top link and put it space at the top it's not very different from what they've already done before
bard came out um i believe bard still hasn't solved the fake news problem completely um it does hallucinate and it doesn't actually have like real citations sometimes um um so we we are better than bard in that in the context of search
we're hiring for ios and android mainly right now so ios engineers if you want to come and help build our mobile experiences please join us
we just want regular daily users to stop using google and use our product
what we offer in the paid version is unlimited uses of that not full like technically unlimited so more like 300 queries a day which is practically unlimited for most people um and then everything else is free
the copilot is just meant to be like a companion like that the world uses the word is a companion for search and it's going to help you plan travel by products um prepare meal plans according to your preferences and if you integrate your ai profile with it it's going to give you like much more detailed recommendations
more like an interactive search companion ah uh that it does the equivalent of uh hundreds of search queries for you not just one so you can ask it really complex queries... it's almost like you're having a researcher at your disposal
yeah so we we don't waste time trying to hire people that same woman will be hiring anyway uh it's very you cannot compete they have way more cash way more like and they can give way less percent of the company because they have way bigger valuation so what we do is go for these people who are still trying to get into ai very talented engineers who haven't done ai before and want to be part of an amazing product that's growing and they want to feel the dopamine from shipping every week and want
i'm more with you here okay so firstly i think relevance can be even more targeted now than ever before i what what is the purpose of google it's just bringing two parties together the advertiser and the consumer and they help you connect these two parties together with their query and link matching right at the end of the day the advertiser wants to get their content to the consumer or the content and llm can give you that needles in the haystack even better like it's even more targeted honestl
ideally if there's a need for us to pay any party for their data access we'll do it
as for how we do it like what links we know for you to use for which query we we do like take your query and figure out like which category it is and like try to use that information to um give you the right sources it's pretty hard actually google does a tremendous job at this and um we are also doing something called focus searches where in the search bar instead of using
it is the case today uh the reason is they have the best models and there's still a lot of differentiation you can have and how you harness the power of these models these are so general purpose machines it's almost like you buy the engine from somewhere but you're building um a whole car with a lot of different parts and you can still build a better car
people have done comparisons between us and bing and um you know we have like much better accuracy in terms of how correctly we cite things a lot of academic research has been done there uh people spend on an average like two minutes more on our site than bing uh so that engagement is much better there so our bounce rates are much lower than bing
there's only one way to win against the person who has much more distribution than you which is a superior product
and i think at some point you're just going to take answers as the de facto search experience that's kind of what we want to bring together
yeah so firstly we uh we started off a week after chat gpt came out we put it out and there was a lot of difference at that time which is we were just a search bar and we give direct answers with citations uh whereas chat gpt was this entertaining hallucinatory bot that was not not just you know like correct many times but it was also equally wrong many times and its mistakes were also entertaining right so we focused a lot more on revamping search realizing that you know 10 years from now no on