[leereeves]

Is it?

There are many mentions of confetti cannons on the web, along with explanations of how they work (saying something like confetti shoots out of the cannon). Chat-GPT just picked a random thing (confetti) and completed the pattern "X out of Y" with the thing confetti comes out of.

It's easy. The cereal is out of the box. The helium is out of the balloon. The snow is out of the globe.

And it's exactly the one thing LLMs are trained to do: reproduce patterns of words. They are (perhaps) already better than humans at that one specific skill (another win for AI research) but I don't think it's a sign of general intelligence.

[slowmovintarget]

Like you, I thought these pieces of software and data were little more than statistics-based text generators. But it turns out that this is a Category Mistake.

There was an argument made by Raphaël Millière in a recent Mindscape Podcast [1] with Sean Carroll that finally landed for me. He used the example that human beings are driven to eat and reproduce, so by that argument all humans are just eating and reproducing machines. "Ah! But we developed other capabilities along the way to allow us to be good at that!" And that's the point.

GPT-4, for example, is very very good at producing pleasing and useful output for a given input. It uses a simulated neural net to do that. Why would one assume that on the way toward becoming excellent at that task that a neural net wouldn't also acquire other abilities that we associate with reasoning or cognition? When we test GPT-4 for these things (like Theory of Mind) we actually find them.

"Ah hah!" you say, "Humans are set up to learn from the get go, and machines must be trained from scratch." However if you consider the entirety of our genetic legacy together with our childhoods, those are our equivalent "training" from scratch.

I don't think it can be easily dismissed that we're seeing something significant here. It's not human-level intelligence yet. Part of the reason for that is that human brains are vastly more complex than any LLM at the moment (100s of trillions of "parameters" in LLM-speak, along with other advantages). But we're seeing the emergence of something important.

[1] https://www.youtube.com/watch?v=aUJOcVPdDvg

[leereeves]

That emergence is precisely what I'm looking for evidence of.

Human beings evolved to eat and reproduce and yet here we are, building computers and inventing complex mathematical models of language and debating whether they're intelligent.

We're so far from the environment we evolved to solve that we've clearly demonstrated the ability to adapt.

ChatGPT doing well at a language task isn't demonstrating that same ability to adapt because that's the task it was designed and trained to do. ChatGPT doing something completely different would be the impressive example.

In short: I don't categorically reject the possibility that LLMs might become capable of more than being "statistics-based text generators", I simply require evidence.

[patrickas]

This paper from late last year shows that LLMs are not "just" stochastic parrots, but they actually build an internal model of the "world" that is not programmed in, just from trying to predict the next token.

https://ar5iv.labs.arxiv.org/html/2210.13382

PS: More research has been done since that confirmed and strengthened the conclusion.

[slowmovintarget]

Another poster shared a link to this paper last week for Theory of Mind: https://arxiv.org/abs/2302.02083

We're seeing those other capabilities emerge; like being able to play chess though it's not been trained to do so. That is, these LLMs are displaying emergent abilities associated with reasoning.

These LLMs aren't R. Daneel Olivaw or R2D2 (which is what I think of when I think of the original term for AI, and what we took to calling AGI). We're closer to seeing the just-the-facts AIs we encounter in Blindsight. Intelligence without awareness.

Funny that we still have to use science fiction to make our comparisons because our philosophy of intelligence, mind, and consciousness are insufficient to speak on the matter clearly.

[roenxi]

The issue I see here is you are doing a worse job at this than ChatGTP. Creating idioms is hard, that is why we left most of them to Shakespeare.

- I regularly return cereal to its box.

- "helium" and "balloon" have a more awkward rhythm than "confetti" and "cannon". It also loses the connotations of sudden, explosive and exciting change.

- Snow & globe I'm not even sure what that means in practice. It has poor prospects as an idiom. Is the snow even known for leaving globes?

[pmoriarty]

> "helium" and "balloon" have a more awkward rhythm than "confetti" and "cannon". It also loses the connotations of sudden, explosive and exciting change.

Not only that, but "the confetti has left the cannon" is an alliteration, which makes the phrase even more poetic.

[leereeves]

I actually agree, it's a very good phrase.

But I do think it's cherry picking the most impressive example. I repeated the dialog (and some variations), each time asking for a completely new idiom, and ChatGPT responded with several phrases that aren't new at all:

"The toothpaste is out of the tube"

"The lid has been lifted"

"The secret has been spilled"

"The arrow has left the bow"

[selcuka]

> But I do think it's cherry picking the most impressive example.

Yes, but don't we cherry pick from what humans have said, too? I'm sure there have been many dumb and obvious proverbs that didn't survive.

> ChatGPT responded with several phrases that aren't new at all

Are you using the GPT-4 version of ChatGPT? That's what GP used.

[DoctorOetker]

it also attests to the ramifications, the magnificent difficulty of cleaning confetti, you will keep occasionally finding confetti around the house for months or years

[techdragon]

To continue with some human brain powered pedantry for fun I’d argue that “the snow is out of the globe” is not a valid idiom to suit the “X out of Y” pattern.

We have multiple example pairings of X and Y but the common components are that putting X back in Y would be impossible or extremely difficult, and also that X is in some way meant to no longer be contained inside Y as part of either a desired outcome, it’s normal function or the natural expected and thus inevitable result. Cats want to escape, helium leaks, confetti is expelled to have the desired effect, and toothpaste is squeezed out to use it…

For the snow to come out of a snow globe you have to smash it which is not normal usage, not normally meant to happen, and shouldn’t happen by itself. Making the idiom “The snow is out of the globe” not a proper member of this “family” of idioms. (Also I’m not sure if there’s an agreed upon collective noun for idioms)

[slyall]

"The AI effect occurs when onlookers discount the behavior of an artificial intelligence program by arguing that it is not real intelligence."

https://en.wikipedia.org/wiki/AI_effect

[leereeves]

Better that than the opposite effect, to assume that because a system solves a single problem very well, it is intelligent.

Is Stockfish intelligent?

Is a system with A* pathfinding intelligent?

I would define intelligence as the ability to solve a wide variety of novel problems. A system built to be excellent at a single task may be better than humans at that task but still lack "intelligence".

We still don't know what that even is exactly, but historically people consistently underestimate how difficult it is.

Not even knowing how to approach it, researchers work on solving single specialized problems instead and make little progress on whatever "intelligence" is.

(If you'd prefer a different definition of intelligence under which Stockfish and GPT are intelligent, then what would you call the ability to solve a wide variety of novel problems? Feel free to substitute that word for "intelligence" above if you'd like to understand what I'm saying.)

[pmoriarty]

> Is Stockfish intelligent?

> Is a system with A pathfinding intelligent?*

I'm not sure if we should get stuck on definitions of intelligence.

The fact is that these tools are useful, as are the currently existing AI's. The latter can also pass for humans, in many ways, while the algorithms you mentioned can only pass for humans in very narrow domains. Both can exceed human performance in some ways.

Eventually, AI's may be indistinguishable from human or convince humans that they should be treated differently from "mere" programs and algorithms, and at that point we will have entered a new era, call it what you will.

[leereeves]

We'll have to "get stuck on" definitions of intelligence if we want to talk precisely about what LLMs are capable of.

[pmoriarty]

Not necessarily, as we can just evaluate them on their performance as we give them ever greater challenges.

To do this we do not need to consider whether they're intelligent at all.

[Timon3]

In a way, this comment perfectly encapsulates why the argument "machines will never replicate human behavior" is so ridiculous. Instead of engaging with the discussion and topic, you chose a position, and then tried to justify it without really thinking about why one example works and the other one doesn't. In doing so you're literally showing that for certain topics, machines are already more capable than some humans.

[leereeves]

I didn't say "machines will never replicate human behavior" so I don't think you're engaging with what I said.

[Timon3]

Sure, you didn't say that. You merely said:

> And it's exactly the one thing LLMs are trained to do: reproduce patterns of words. They are (perhaps) already better than humans at that one specific skill (another win for AI research) but I don't think it's a sign of general intelligence.

while completely missing why the machine did a better job.