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

Using secondary school maths to demystify AI

raspberrypi.org

Post reply on HN
Print view View original

51–60 of 264 posts

Re: Using secondary school maths to demystify AI

#51

Post by pegasus »

snickerbockers wrote: post #23
causal wrote: post #2

Earlier quoted context omitted.

I love the idea of educating students on the math behind AI to demystify them. But I think it's a little weird to assert "AI is not magic and AI systems do not think. It’s just maths." Equivalent statements could be made about how human brains are not magic, just biology - yet I think we still think.

>Equivalent statements could be made about how human brains are not magic, just biology - yet I think we still think. They're not equivalent at all because the AI is by no means biological. "It's just maths" could maybe be applied to humans but this is backed entirely by supposition and would ultimately just be an assumption of its own conclusion - that human brains work on the same underlying principles as AI becaus…

But parent didn't try to apply "it's just maths" to humans. He said one could just as easily say, as some do: "Humans are just biology, hence they're not magic". Our understanding of mathematics, including the maths of transformer models is limited, just as our understanding of biology. Some behaviors of these models have taken researches by surprise, and future surprises are not at all excluded. We don't know exactly how far they will evolve.

As for applying the word thinking to AI systems, it's already in common usage and this won't change. We don't have any other candidate words, and this one is the closest existing word for referencing a computational process which, one must admit, is in many ways (but definitely not in all ways) analogous to human thought.

Re: Using secondary school maths to demystify AI

#52

Post by hnfong »

palmotea wrote: post #21
School-Cotton wrote: post #8

Earlier quoted context omitted.

You can simulate a human brain on pen and paper too.

> You can simulate a human brain on pen and paper too. That's an assumption, though. A plausible assumption, but still an assumption. We know you can execute an LLM on pen and paper, because people built them and they're understood well enough that we could list the calculations you'd need to do. We don't know enough about the human brain to create a similar list, so I don't think you can reasonably make a stronger s…

This is basically the Church-Turing thesis and one of the motivations of using tape(paper) and an arbitrary alphabet in the Turing machine model.

It's been kinda discussed to oblivion in the last century, interesting that it seems people don't realize the "existing literature" and repeat the same arguments (not saying anyone is wrong).

Re: Using secondary school maths to demystify AI

#53

Post by nh23423fefe »

CamperBob2 wrote: post #40
Terr_ wrote: post #36

Earlier quoted context omitted.

Someone asserts, almost religiously, that LLMs do and/or can "think." When asked how to falsify their assertion, perhaps by explaining what exactly is "thinking" in the human brain that can and/or will be possible to emulate...

When asked how to falsify their assertion, perhaps by explaining what exactly is "thinking" in the human brain that can and/or will be possible to emulate... ... someone else points out that the same models that can't "think" are somehow turning in gold-level performance at international math and programming competitions, making Fields Medalists sit up and take notice, winning art competitions, composing music indist…

god of the gaps

Re: Using secondary school maths to demystify AI

#54

Post by CamperBob2 »

Terr_ wrote: post #42
CamperBob2 wrote: post #40

Earlier quoted context omitted.

When asked how to falsify their assertion, perhaps by explaining what exactly is "thinking" in the human brain that can and/or will be possible to emulate... ... someone else points out that the same models that can't "think" are somehow turning in gold-level performance at international math and programming competitions, making Fields Medalists sit up and take notice, winning art competitions, composing music indist…

A couple decades of chess programs nods knowingly: "First time?"

A couple decades of chess programs nods knowingly: "First time?"

Uh huh. Good luck getting Stockfish to do your math homework while Leela works on your next waifu.

LLMs play chess poorly. Chess engines do nothing else at all. That's kind of a big difference, wouldn't you say?

Re: Using secondary school maths to demystify AI

#55

Post by superkuh »

Terr_ wrote: post #36
CamperBob2 wrote: post #29

Earlier quoted context omitted.

That's where these threads always end up. Someone asserts, almost violently, that AI does not and/or cannot "think." When asked how to falsify their assertion, perhaps by explaining what exactly is unique about the human brain that cannot and/or will not be possible to emulate, that's the last anyone ever hears from them. At least until the next "AI can't think" story gets posted. The same arguments that appeared in…

Someone asserts, almost religiously, that LLMs do and/or can "think." When asked how to falsify their assertion, perhaps by explaining what exactly is "thinking" in the human brain that can and/or will be possible to emulate...

Or they just point to the turing test which was the defacto standard test for something so nebulous. And behold: LLM can pass the turing test. So they think. Can you come up with something better (than the turing test)?

Re: Using secondary school maths to demystify AI

#56

Post by terminalshort »

palmotea wrote: post #21
School-Cotton wrote: post #8

Earlier quoted context omitted.

You can simulate a human brain on pen and paper too.

> You can simulate a human brain on pen and paper too. That's an assumption, though. A plausible assumption, but still an assumption. We know you can execute an LLM on pen and paper, because people built them and they're understood well enough that we could list the calculations you'd need to do. We don't know enough about the human brain to create a similar list, so I don't think you can reasonably make a stronger s…

I can make a claim much stronger than "you could probably" The counterclaim here is that the brain may not obey physical laws that can be described by mathematics. This is a "5G causes covid" level claim. The overwhelming burden of proof is on you.

Re: Using secondary school maths to demystify AI

#57

Post by ehsanu1 »

phantasmish wrote: post #15
space_fountain wrote: post #13

Earlier quoted context omitted.

It’s not that open. We can simulate smaller system of neurons just fine, we can simulate chemistry. There might be something beyond that in our brains for some reason, but it sees doubtful right now

Our brains actually do something , may be the difference. They're a thing happening, not a description of a thing happening. Whatever that something that it actually does in the real, physical world is produces the cogito in cogito, ergo sum and I doubt you can get it just by describing what all the subatomic particles are doing, any more than a computer or pen-and-paper simulated hurricane can knock your house down,…

Doing something merely requires I/O. Brains wouldn't be doing much without that. A sufficiently accurate simulation of a fundamentally computational process is really just the same process.

Re: Using secondary school maths to demystify AI

#58

Post by nutjob2 »

CamperBob2 wrote: post #40
Terr_ wrote: post #36

Earlier quoted context omitted.

Someone asserts, almost religiously, that LLMs do and/or can "think." When asked how to falsify their assertion, perhaps by explaining what exactly is "thinking" in the human brain that can and/or will be possible to emulate...

When asked how to falsify their assertion, perhaps by explaining what exactly is "thinking" in the human brain that can and/or will be possible to emulate... ... someone else points out that the same models that can't "think" are somehow turning in gold-level performance at international math and programming competitions, making Fields Medalists sit up and take notice, winning art competitions, composing music indist…

Computers can perform math and numerous other tasks billions of times faster than humans, whats your point?

This is exactly the problem. Claims about AI are unfalsifiable, thus your various non-sequiturs about AI 'thinking'.

Re: Using secondary school maths to demystify AI

#59

Post by sigmoid10 »

gus_massa wrote: post #27
frozenlettuce wrote: post #6

Earlier quoted context omitted.

You can replicate all calculations done by LLMs with pen and paper. It would take ages to calculate anything, but it's possible. I don't think that pen and paper will ever "think", regardless of how complex the calculations involved are.

The official name is https://en.wikipedia.org/wiki/Chinese_room The opinions are exactly the same than about LLM.

And the counter argument is also exactly the same. Imagine you take one neuron from a brain and replace it with an artificial piece of electronics (e.g. some transistors) that only generates specific outputs based on inputs, exactly like the neuron does. Now replace another neuron. And another. Eventually, you will have the entire brain replaced with a huge set of fundamentally super simple transistors. I.e. a computer. If you believe that consciousness or the ability to think disappears somewhere during this process, then you are essentially believing in some religious meta-physics or soul-like component in our brains that can not be measured. But if it can not be measured, it fundamentally can not affect you in any way. So it doesn't matter for the experiment in the end, because the outcome would be exactly the same. The only reason you might think that you are conscious and the computer is not is because you believe so. But to an outsider observer, belief is all it is. Basically religion.

Re: Using secondary school maths to demystify AI

#60

Post by CamperBob2 »

nh23423fefe wrote: post #53
CamperBob2 wrote: post #40

Earlier quoted context omitted.

When asked how to falsify their assertion, perhaps by explaining what exactly is "thinking" in the human brain that can and/or will be possible to emulate... ... someone else points out that the same models that can't "think" are somehow turning in gold-level performance at international math and programming competitions, making Fields Medalists sit up and take notice, winning art competitions, composing music indist…

god of the gaps

Exactly. As soon as a model does something it "wasn't supposed to be able to do," two gaps open up on either side.
Post reply on HN