—
no predictions resolved yet
0 pending · 0 expired
You know, we didn't know how well this would work. This actually works rather well. I mean, we have very conveniently, we have, I mean, in the technicalities of the plugin, it has two different endpoints inside it. One of them is going to Wolfram Alpha. Wolfram Alpha takes natural language input, takes small fragments of natural language. And Wolfram language is a precise computational language. And sometimes what ChatGPT is doing is taking this big lump of text that somebody might have given as
the plugin that we just worked on with OpenAI, you know, the Wolfram plugin for ChatGPT, that's what it's achieving, is being able to connect this kind of LLM layer to this sort of what we might think of as kind of computational bedrock of what one can compute from. And, you know, that's, there's all sorts of implications.
people ask, for example, does ChatGPT understand what it's talking about? Well, it just has these rules that say how the next word goes in, it doesn't, you could, I mean, that's how we work to probably. And you can ask, do we understand what we're talking about, so to speak? And there isn't this, but but it is in a sense, doing a very shallow computation
suddenly it was exciting, and it was able to do really useful things.