[night862]

Well, it isn’t great. Computers are logic, and LLM tools are not really able to derive logic from language.

Is there is enough logic q&a yeah they manage to produce something like what people talk about.

My experience with this has been terrible, I find them useless.

[kragen]

monsieur, what good is a new-born baby?

[rockskon]

LLMs are not intellence. They are not AGI. They do not have the reasoning skills of a baby. This is not something they will "grow into". No amount of additional training data in the world will turn LLMs into AGI. Any argument to the contrary is pure, 100% copium to justify the hundreds of billions of dollars industry has poured into this technology.

[kragen]

hot air balloons aren't birds either, monsieur

[rockskon]

That analogy still doesn't mean LLMs are a form of intelligence with reasoning capability.

[kragen]

it also doesn't mean sulfur is a metal

[night862]

If and when you try to derive logic from the statistical model that is an LLM, you will face the same problem that Bertrand Russel did.

Good luck.