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

AI capability isn't humanness

research.roundtable.ai

Post reply on HN
Print view View original

31–40 of 55 posts

Re: AI capability isn't humanness

#31

Post by danaris »

ACCount37 wrote: post #12
gmuslera wrote: post #10

LLMs are language models. We interact with them using language, all of that, but also only that. That doesn't mean that they have "common sense", context, same motivations, agency, or even reasoning like us. But as we interact with other people using mostly language, and since the start of internet a lot of those interactions happen in way similar to how we interact with AI, the difference is not so obvious. We are f…

"Language" is just the interface. What happens on the inside of LLMs is a lot weirder than that.

"Weirder" does not mean "more complex" or "more human-like".

Re: AI capability isn't humanness

#32

Post by measurablefunc »

ACCount37 wrote: post #12
gmuslera wrote: post #10

LLMs are language models. We interact with them using language, all of that, but also only that. That doesn't mean that they have "common sense", context, same motivations, agency, or even reasoning like us. But as we interact with other people using mostly language, and since the start of internet a lot of those interactions happen in way similar to how we interact with AI, the difference is not so obvious. We are f…

"Language" is just the interface. What happens on the inside of LLMs is a lot weirder than that.

Which arithmetic operation in an LLM is weird?

Re: AI capability isn't humanness

#33

Post by ACCount37 »

layer8 wrote: post #17
cortesoft wrote: post #9

Earlier quoted context omitted.

Not only that, but humans also have access to all of the "training data" of hundreds of millions of years of evolution baked into our brains.

I don’t think the amount of data is essential here. The human genome is only around 750 MB, much less than current LLMs, and likely only a small fraction of it determines human intelligence. On the other hand, current LLMs contain immense amounts of factual knowledge that a human newborn carries zero information about. Intelligence likely doesn’t require that much data, and it may be more a question of evolutionary c…

That just says "low Kolmogorov complexity". All the priors humans ship with can be represented as a relatively compact algorithm.

Which gives us no information on computational complexity of running that algorithm, or on what it does exactly. Only that it's small.

LLMs don't get that algorithm, so they have to discover certain things the hard way.

Re: AI capability isn't humanness

#34

Post by omneity »

mdahardy wrote: post #21
ForceBru wrote: post #5

> Compared to humans, LLMs have effectively unbounded training data. They are trained on billions of text examples covering countless topics, styles, and domains. Their exposure is far broader and more uniform than any human's, and not filtered through lived experience or survival needs. I think it's the other way round: humans have effectively unbounded training data. We can count exactly how much text any given mod…

This is a fair criticism we should've addressed. There's actually a nice study on this: Vong et al. ( https://www.science.org/doi/10.1126/science.adi1374 ) hooked up a camera to a baby's head so it would get all the input data a baby gets. A model trained on this data learned some things babies do (eg word-object mappings), but not everything. However, this model couldn't actively manipulate the world in the way that…

While an LLM is trained on trillions of tokens to acquire its capabilities, it does not actively retain or recall the vast majority of it, and often enough is not able to make deductive reasoning either (e.g. X owns Y does not necessarily translate to Y belongs to X).

The acquired knowledge is a lot less uniform than you’re proposing and in fact is full of gaps a human would never make. And more critically, it is not able to peer into all of its vast knowledge at once, so with every prompt what you get is closer to an “instance of a human” than “all of humanity” as you might think of LLMs.

(I train and dissect LLMs for a living and for fun)

Re: AI capability isn't humanness

#35

Post by ACCount37 »

measurablefunc wrote: post #32
ACCount37 wrote: post #12

Earlier quoted context omitted.

"Language" is just the interface. What happens on the inside of LLMs is a lot weirder than that.

Which arithmetic operation in an LLM is weird?

The fact that you can represent abstract thinking as a big old bag of matrix math sure is.

Re: AI capability isn't humanness

#37

Post by ACCount37 »

gmuslera wrote: post #29
ACCount37 wrote: post #12

Earlier quoted context omitted.

"Language" is just the interface. What happens on the inside of LLMs is a lot weirder than that.

What matters is what happen in the outside. We don't know what happen in our inside (or the inside of others, at least), we know the language and how it is used, event the meanings don't have to be the same as long as it is consistent. And you get that by construction. Does that mean intelligence, self consciousness, soul or whatever? We only know that it walk like a duck and quacks like a duck.

But have you considered that humans really really want to feel like they're unique and special and exceptional?

If it walks like a duck and quacks like a duck, then it's not anything remotely resembling a duck, and is just a bag of statistics doing surface level duck imitation. According to... ducks, mostly.

Re: AI capability isn't humanness

#38

Post by minraws »

omneity wrote: post #34
mdahardy wrote: post #21

Earlier quoted context omitted.

This is a fair criticism we should've addressed. There's actually a nice study on this: Vong et al. ( https://www.science.org/doi/10.1126/science.adi1374 ) hooked up a camera to a baby's head so it would get all the input data a baby gets. A model trained on this data learned some things babies do (eg word-object mappings), but not everything. However, this model couldn't actively manipulate the world in the way that…

While an LLM is trained on trillions of tokens to acquire its capabilities, it does not actively retain or recall the vast majority of it, and often enough is not able to make deductive reasoning either (e.g. X owns Y does not necessarily translate to Y belongs to X). The acquired knowledge is a lot less uniform than you’re proposing and in fact is full of gaps a human would never make. And more critically, it is not…

I think you are proposing something that's orthogonal to the OP's point.

They mentioned the training data is much higher for an LLM, LLM's recall not being uniform was never in question.

No one expects compression to be without loss when you scale below knowledge entropy that exists in your training set.

I am not saying LLMs do simple compression but just pointing a mathematical certainity.

(And I think you don't need to be an expert in creating LLMs to understand them, albeit I think a lot of people here have experience with it aswell so I find the additional emphasis on it moot).

Re: AI capability isn't humanness

#39

Post by ACCount37 »

measurablefunc wrote: post #36
ACCount37 wrote: post #35

Earlier quoted context omitted.

The fact that you can represent abstract thinking as a big old bag of matrix math sure is.

So it's not weird, it's actually very mundane.

If your takeaway from the LLM breakthrough is "abstract thinking is actually quite mundane", then at least you're heading in the right direction. Some people are straight up in denial.

Re: AI capability isn't humanness

#40

Post by hathawsh »

ACCount37 wrote: post #11
cortesoft wrote: post #9

Earlier quoted context omitted.

Not only that, but humans also have access to all of the "training data" of hundreds of millions of years of evolution baked into our brains.

Which must be doing some heavy lifting. Humans ship with all the priors evolution has managed to cram into them. LLMs have to rediscover all of it from scratch just by looking at an awful lot of data.

OTOH, all that data is built on patterns that evolved from many years of evolution, so I think the LLM benefits from that evolution also.
Post reply on HN