Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A Multi-Level View of LLM Intentionality

disagreeableme.blogspot.com

Post reply on HN
Print view View original

71–75 of 75 posts

Re: A Multi-Level View of LLM Intentionality

#71

Post by passion__desire »

pyinstallwoes wrote: post #67
passion__desire wrote: post #63

Earlier quoted context omitted.

So, the reason why a Boltzmann brain would vanish almost instantly is because the longer it is to survive the more support structures it would need, a larger part of the local environment would have to be compatible with a longer existence, and all that makes it less likely for the fluctuation yielding such an outcome to have occurred.

What's time to a boltzmann brain?

I don't remember the video, but it was pointed out by Sean Carroll that the idea of BB is not fleshed out. Does one need only neo-cortex, or full brain, or just few control centers of attention within neo-cortex to be called as BB.

Re: A Multi-Level View of LLM Intentionality

#73

Post by bt1a »

astrange wrote: post #37
famouswaffles wrote: post #15

Earlier quoted context omitted.

>I do wonder though if we give the LLMs enough examples of texts with people describing their relative spatial position to each other and things will it eventually "learn" to work things these out a bit better GPT-4's spatial position understanding is actually really good all things considered. By the end, 4 was able to construct an accurate maze just from feedback about the current position and possible next moves a…

A funny thing GPT-4 is unusually good at is giving driving directions. This shouldn't work, and of course isn't 100% right, but… it's kind of right. Bard can answer questions like this, but I think it actually uses the Maps API. (It certainly says that's what it's doing.) On the other hand, every chatbot including GPT-4 is both unable to do ASCII art and unable to tell it can't do it. (Bard always shows you `cowsay`…

I tried so hard to make ascii art with GPT-4 api :(

Re: A Multi-Level View of LLM Intentionality

#74

Post by gwd »

My analogy for GPT-4 is this: GPT-4 is writing a novel, in which a human talks to a very smart AI. This helps me contextualize its hallucinations: if I were writing such a novel and I knew the answer to something, I would put in the correct answer; if I were writing such a novel and I didn't know the answer to something (and had no way to look it up), I would make up something plausible.

From that perspective, I think multi-intentionality also works. If I write a story about Bob, then Bob (in the story) has intentions, although he's just a figment of my imagination; and when we read characters in novels, we use the imputed intentions of the characters to understand their behavior, although we know they're fictional and don't actually exist.

So yes; on one level, I want to write an exciting story; on a second level, I'm simulating Bob in my head, who wants to execute the perfect robbery. On one level, GPT-4 wants to write a story about a smart AI; on a second level, the smart AI in GPT-4's story wants to win the chess game by moving the queen to put the king in check.

Re: A Multi-Level View of LLM Intentionality

#75

Post by mgraczyk »

The reason why it isn't useful to ascribe intentions without a mechanistic explanation of intentionality is because you will incorrectly predict what the model will do in surprising ways.

I think it's true that current generation LLMs could, in principle, have intentionality in the way described in the article. But they would have to be trained on many orders of magnitude more data than current models.

Also AutoGPT does not work. I encourage the author to play around with it and try to get it to do something useful with high success probability.

Post reply on HN