3 ms·

Years ago there was an idea called "the semantic web": https://en.wikipedia.org/wiki/Semantic_Web https://en.wikipedia.org/wiki/Semantic_Web The basic idea wa

Years ago there was an idea called "the semantic web": https://en.wikipedia.org/wiki/Semantic_Web https://en.wikipedia.org/wiki/Semantic_Web The basic idea was to have enough metadata about web sites so that you could get programs to do something approaching Prolog-style reasoning about the content and meaning of the web pages. With more advanced LLMs it looks like a slightly different approach to achive something like the semantic web idea. I think the idea is to constantly feed the model with updates from crawling the web and have the LLM "digest" the content, apply some filters to remove bad stuff, and then provide a meaningful result to whatever queries it might be asked.

PaulHoule 4y ago
I'd picture that as a hybrid system that uses some old AI ideas together with some new AI ideas. AlphaGo which combines a neural network-based player with a MCMS player that simulates a large number of games all the way to the end, is a good example. The intellectual north star of this is https://en.wikipedia.org/wiki/SAT_solver https://en.wikipedia.org/wiki/SAT_solver and might be a better model than the "semantic web" in that the semantic web has revolved around description logics like OWL that implement a limited subset of first-order logic that is always decidable. One trouble is that OWL-based systems can't handle arithmetic at all, given a height and a weight you can write a rule that says anybody over 6 feet is tall but you can't say a person with a BMI over 30 is obese because OWL can't add, multiply and divide. But something a little more expressive that searches efficiently in practice because it gets "hunches" from a neural net could be useful but it will not evade the limits on computation and logic that Godel, Tarski and Turing warned you about.
- syntaxfree 4y ago
  SAT solvers <3 there’s something about SATPLAN that feels like it has metaphysical significance.
dboreham 4y ago
> Years ago there was an idea called... That idea was the basis for the current generation of Google search (see: Metaweb).

PaulHoule 4y ago

I'd picture that as a hybrid system that uses some old AI ideas together with some new AI ideas. AlphaGo which combines a neural network-based player with a MCMS player that simulates a large number of games all the way to the end, is a good example. The intellectual north star of this is https://en.wikipedia.org/wiki/SAT_solver https://en.wikipedia.org/wiki/SAT_solver and might be a better model than the "semantic web" in that the semantic web has revolved around description logics like OWL that implement a limited subset of first-order logic that is always decidable. One trouble is that OWL-based systems can't handle arithmetic at all, given a height and a weight you can write a rule that says anybody over 6 feet is tall but you can't say a person with a BMI over 30 is obese because OWL can't add, multiply and divide. But something a little more expressive that searches efficiently in practice because it gets "hunches" from a neural net could be useful but it will not evade the limits on computation and logic that Godel, Tarski and Turing warned you about.

syntaxfree 4y ago
SAT solvers <3 there’s something about SATPLAN that feels like it has metaphysical significance.

dboreham 4y ago

> Years ago there was an idea called... That idea was the basis for the current generation of Google search (see: Metaweb).