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

And yet It Understands

borretti.me

Post reply on HN
Print view View original

201–210 of 231 posts

Re: And yet It Understands

#201

Post by lurquer »

datatrashfire wrote: post #158
glenstein wrote: post #65

Earlier quoted context omitted.

>means that even chairs and rockets share some attributes (helping you get higher), the difference between them is still qualitative, not quantitve. I don't think so, because the needed 'quality' is the ability to traverse space. So I don't think I agree that the qualitative piece is missing. Perhaps the moon example is helpful here because the real solution, a rocket ship, uses propulsion rather than sheer mass, and…

They are qualitatively different because the rocket can traverse that amount of space scalably. The chairs cannot.

Of course they can. Neal Armstrong literally travelled to the moon sitting in a literal chair.

Basically, the Saturn rocket was more or less a simple chair with some extra features added to it.

Granted they didn’t stack chairs. But, nevertheless, focusing on the mods you’d need to make to your La-z-boy to get to moon would be a fruitful way of approaching the problem.

Re: And yet It Understands

#202

Post by marcosdumay »

YeGoblynQueenne wrote: post #164
marcosdumay wrote: post #133

Earlier quoted context omitted.

If it stopped making mistakes and produced complete fully working programs, no, there would be no way to say it doesn't understand. Yes, "understanding" is a process, but it's not well defined. And anyway, if it's a requirement for those things, and the AI did those things, then the only possibility is that the AI has this process in some way. But well, our current AIs do not produce complete programs, nor fully work…

>> If it stopped making mistakes and produced complete fully working programs, no, there would be no way to say it doesn't understand. There exist program synthesis systems that always produce "complete fully working programs" and that in fact cannot make any mistakes because they are based on algorithms with strong theoretical guarantees (like mathematical proofs) of their correctness. You give them a set of example…

> a program synthesis system that never makes mistakes "understands"

Yes, I'm saying if it satisfies random requirements and synthesizes full working programs, then it has some form of understanding in it. If creates that understanding through learning, it's very possibly a general form of AI (I'm not certain of this part).

I've never seen any one that arrived at structure by learning. It's always programed in, what puts the intelligence on a very conventional place, and I doubt you'll disagree that it understands what it's doing.

Re: And yet It Understands

#203

Post by sdifnhiono »

ccppurcell wrote: post #8

I just asked chatgpt whether 3442177452 is prime. It insisted that 58657 is a factor (it's not) on the basis that it's the largest prime less than or equal to the square root (which I think is correct but irrelevant), and even though it gave a non zero remainder when dividing the two numbers (I did not check if the remainder is correct). Then it gave a (wrong) factorisation, not even using 58657. It's cool and it wil…

These LLMs are trained to predict text. "Fancy autocomplete" is not just a dismissal. It is the fundamental nature of these things. It is an autocomplete tool hacked together to emulate an assistant. "I don't know" is clearly a truer answer than the nonsense ChatGPT made up, but that doesn't mean it's a better answer according to whatever non-goals this machine has.

As always, anthropomorphize an LLM at your peril. These things are not people. No one understands how they work.

Re: And yet It Understands

#204

Post by sdifnhiono »

_dain_ wrote: post #92
ivanbakel wrote: post #72

Earlier quoted context omitted.

>I don't see how it can do that unless it really has some kind of understanding. One possibility is that the model itself has learned that tokens are related across languages based on translation examples. If the appended training changes the model's treatment of tokens in one language, that could have a statistical knock-on effect on the weights between similar tokens in different languages. Similarly, if you train…

At the scale/complexity GPT operates at, how is that different from "understanding"? It seems like you've just rephrased it but with more words.

The fact that something is hard to define does not mean it isn't meaningful. I cannot tell you what consciousness is, but I know that I have it and my sock does not.

It is trivial to tie these machines up in knots, blatantly contradicting themselves from one sentence to the next. If that is understanding then it is an understanding utterly foreign to any human, and a form of understanding that looks a whole lot like bullshit.

Re: And yet It Understands

#206

Post by branko_d »

smitty1e wrote: post #124
branko_d wrote: post #121

Earlier quoted context omitted.

> if our brains are so sophisticated that they make natural intelligence possible, then we should be so intelligent that we would have the means to understand them It may well be that building a brain is easier than understanding it.

If one has not understood the brain, then how would one be certain that such has been built?

The same way we do it with humans - by observing the end result.

Re: And yet It Understands

#207

Post by meghan_rain »

freehorse wrote: post #60
Etheryte wrote: post #11

Earlier quoted context omitted.

This is a common misconception. ChatGPT is not supposed to be good at this — it's a language model, not a maths model or data science model or whatnot. This is exactly why they have plugins, such as the one for Wolfram Alpha.

Interfacing with APIs does not solve any fundamental issues about what LLMs "understand" or not though. At most they will be more accurate in arithmetic tasks, but that's it.

Did you "understand" that 58657 is/is not a factor of 3442177452?

Or did you use a calculator to verify?

Because that would be exactly what ChatGPT with plugins does.

Re: And yet It Understands

#209

Post by yamrzou »

perryizgr8 wrote: post #197
yamrzou wrote: post #46

I said it here before and I will repeat it: Unless it solves the Abstraction and Reasoning Corpus — ARC (See: https://twitter.com/fchollet/status/1636054491480088823 ) you can not say that ChatGPT is able to think or abstract.

Wow that's a really high bar to clear! I consider myself to be a non-dumb person and it would take genuine concentrated thought to figure out the task in the example image. But I think even if GPT-X solves it some people will say it's just regurgitating whatever words and images and associations it has seen in training. There was a time when natural language conversation was considered the gold standard for AI. Now i…

> some people will say it's just regurgitating whatever words and images and associations it has seen in training

That's why I specifically mentioned ARC: Its test test is novel and fully private (even to us, humans), so the model will need abstraction capabilities in order to solve it.

Re: And yet It Understands

#210

Post by ccppurcell »

Etheryte wrote: post #11
ccppurcell wrote: post #8

I just asked chatgpt whether 3442177452 is prime. It insisted that 58657 is a factor (it's not) on the basis that it's the largest prime less than or equal to the square root (which I think is correct but irrelevant), and even though it gave a non zero remainder when dividing the two numbers (I did not check if the remainder is correct). Then it gave a (wrong) factorisation, not even using 58657. It's cool and it wil…

This is a common misconception. ChatGPT is not supposed to be good at this — it's a language model, not a maths model or data science model or whatnot. This is exactly why they have plugins, such as the one for Wolfram Alpha.

I wasn't asking for complex mathematics, just for the basic understanding that an even number greater than 2 can't be prime.
Post reply on HN