[GalahiSimtam]

Yeah it lies, indeed, "print(list(filter(lambda x:all(x%d for d in range(2,x)), range(2,3*10)))[:11])" will print the list of first eleven primes, up to including 31.

So if you want to say that state-of-the-art LLMs could recognize the user intent better than existing voice assistant products, just say that.

I am afraid there's no point in decoupling the problem of lying from that; there is no silver bullet that would fix it

[kilgnad]

What I'm saying is that it's capable of true understanding of intent and subject matter.

It doesn't always understand but many times it does.

[GalahiSimtam]

Though from what I gather about the computation, it does not easily separate into two steps, "understanding"/"generation".

[kilgnad]

Agreed. In fact I think the two steps are remarkably similar in concept and an aspect of what human brains are actually doing.

One thing is for sure though, there are certain responses chatGPT generates that are virtually impossible to "generate" without "understanding" that includes the virtual machine article I linked.