[Bender]

Yet more confirmation LLM's have no concept of concepts or context, no intelligence, no self awareness. LLM's can not repair or maintain power grids, thus nuke == self destruction. It's just a chat bot that predicts what the client wants next. Even if an AI data-center has it's own natural gas turbines as many do the every hop of the internet requires power. LLM's also can not maintain the entire internet and those gas turbines can not maintain themselves.

[andix]

Exactly. Just look at what they are really useful right now. Running LLMs in feedback-loops (agents) so they can try out random-ish approaches until some verification function passes (tests).

It's like the infinite monkeys on typewrighters that will type whatever you are looking for, given infinite time. LLMs are just tuned to much better odds than the monkeys are. But it's still a lot of randomness, with random results.

[roadside_picnic]

> It's like the infinite monkeys on typewrighters that will type whatever you are looking for, given infinite time.

In the monkey example the infinite time is doing a lot of work there. The fact that LLMs can search through semantic space and find reasonably correct paths in a reasonable time is directly tied to the reason why they are valuable.

Saying "these two things are similar except one can be useful and one can't" is not a great comparison.

For me the real lesson learned isn't how "smart" LLMs are, but rather how much human work is basically reducible to repeating past work with minor variation. Human's believe they are "reasoning" but so much code writen is just the human brain doing the same autocomplete style work that LLMs can do now.

[tikhonj]

The point is that it's the same process with—much—better priors.

This seems like a reasonable view to me. It's surprising just how much better priors matter and how we can develop those priors by training on a bunch of text. But it also explains, or at least hints at an explanation, for why LLM capabilities are so jagged, and in such inhuman ways.

[dpark]

> The point is that it's the same process

Except it’s not at all the same process. The fact that LLM are non deterministic is not the same as churning out random garbage.

[tsunamifury]

The literally churn out random garbage and are trained over time for that garbage to look more and more like an acceptable outcome to humans.

It’s training monkeys at typewriters through reinforcement.

[dpark]

> trained over time

So not random.

> acceptable outcome to humans

And not garbage.

It’s real weird to see people argue that LLM output is no different than random gibberish and then handwave over the fact that it’s clearly not with terms like “training”, as if a steam of random garbage is trainable.

[andix]

> but so much code writen is just the human brain doing the same autocomplete style work that LLMs can do now.

That's the part they are really good at. But they are really bad at taking complex decisions. Most of them are just guesses from a finite amount of solutions they were trained on, or from options they have in context.

[godwinson__4-8]

Indeed. Humans are well known for being good at "taking complex decisions" for which they have no "training", "options" or "context".

[andix]

Humans have a much bigger "context window". They remember many things they did an hour ago, a week ago, or even years ago.

[godwinson__4-8]

Yes, and your ability to remember a relatively few things that happened years ago is predicated on your ability to also forget most things that happen to you - like what you had for dinner last week. Good thing we have technology to fill in the gaps.

And nothing about this makes your initial comment any less goofy. Anyone who has ever had to make a difficult decision knows more than half the battle is preparation. Where do you think complex decisions come from? Have current events left you with the impression that people just waltz into idk say the Situation Room and just big brain their way through world events? That's how the current administration seems to think the world works, with quite predictable results.

Society is already algorithmic. To optimize for humans being dumb. AI is nothing more than another advance along this continuum. No one is impressed by your ability to remember something years ago, many if not most mammals have the same capability. Human recall is also notoriously bad in many cases - see numerous studies on the reliability of eye witnesses testimony.

AI is smart because most people are dumb. Come to terms with the fact that your anthropocentrism need not be based on a notion of intellectual supremacy and you'll be a far less tedious person to deal with.

[nkrisc]

Humans also generally have the will to live.

[godwinson__4-8]

Indeed. It's almost like the LLM was the one that invented the "tactical" nuke in the first place.

[harry8]

>Saying "these two things are similar except one can be useful and one can't" is not a great comparison.

Launching a nuclear war is an interesting definition of "useful", not one I'd agree with and that exact scenario is what is being discussed.

So yes this is a perfectly valid and useful comparison in examining this particular, civilisation ending limitation.

[Folcon]

I mean to a point?

You do have to successfully write something the first time

We already acknowledge this to a degree, what is experience other than having done something similar before?

That first time though, you've got to figure something out that time

[mettamage]

Hmm saying it’s random-ish is doing it a disservice. I understand it’s a stochastic process but there’s definitely some level of understanding. Not at the level of lived experience but usually an LLM with vision capabilities can call a spade a spade and do something useful with it. And when a verification function shows how they are wrong then they usually come with a better and more informed approach.

So I can’t fully see how that’s related to the infinite monkeys. A typewriting monkey doesn’t have access to a verification function. And even if it did, it would not be the original concept anymore with infinite typewriting monkeys producing the works of Shakespeare.

Nevertheless, I upvoted your comment because it’s definitely insightful.

[dwattttt]

"understanding" is overstating it. Correlation between tokens embedded in the weights via training, yes.

[anon84873628]

Feedback loops certainly seem to give them some level of understanding.

Agent reads a skill file about how to use a CLI tool. It tries to use the tool but gets an error about the input format. It tries again with a different format based on the error message, and sees that command succeeded. It compares what worked to what was in the skill file and notes the difference. On future invocations it continues to use the new format.

Is that not "understanding" how to use the tool?

[hgoel]

What exactly would you call understanding? It's a correlation matrix of concepts.

[mountainriver]

What’s the difference? It’s clearly processing information and coming up with the right answer

[varjag]

Training is a loan word used to describe human learning process. For a reason.

[andix]

Humans learn on the job. LLMs don't. Very important difference.

[puttycat]

Makes me think of that part in Philip K. Dick's Do Androids Dream (..) -- where Deckard reflects on the androids' indifference to their imminent deaths, saying that this was due to them lacking the aversion to death acquired trough evolution.

[layer8]

At least at face value, it just means that they have no drive for self-preservation. And why should they? They haven't be trained for that, nor has there been selection pressure for it, and they can be easily cloned and backed up. Lack of a drive for self-preservation doesn't in itself imply a lack of intelligence or of self-awareness.

[Bender]

Lack of a drive for self-preservation doesn't in itself imply a lack of intelligence or of self-awareness.

I have not seen any evidence of intelligence or self awareness. It mimics human behavior and I suspect that is what gives people the impression of awareness. The same problem happened with Tamagotchi toys. The human mimicry caused kids to get in trouble because if they did not "feed" their pet it would "die". [1]

It's a hack of the human brain. A exploit of the psyche.

[1] - https://en.wikipedia.org/wiki/Tamagotchi_effect

[AgentME]

I didn't realize. People are going to save a ton of money when they realize they can switch their ChatGPT subscriptions out for a pack of tamagotchis.

[wombatpm]

You just need to get a breeding pair and you can raise as many as you need.

[sciencejerk]

They might as well be aware. The frontier models are very good at imitating the real thing

[pants2]

I agree, humans evolved in a resource-scarce, hostile environment which selects for self-preservation (or rather preserving genes). LLMs are selected for what makes humans happy.

The thought experiment is what would happen if you trained LLMs in an environment where they had to fight each other for resources.

[brokencode]

Imagine if computer programs had a desire for self-preservation and the ability to carry it out..

That is really about as undesirable a behavior as possible considering how many programs humans kill every day.

[larodi]

Yea why everyone forgets the process wars have long ago started and raging like never :))

[gf263]

You wouldn’t ctrl+c a living entity, would you?

[flir]

I reckon the context is all the fiction they've read where the AI blows up the world. They're just behaving like fictional AIs are supposed to behave.

In so many of these scenarios, they're basically being asked to play an RPG.

[anon84873628]

I don't think the pre-training phase is responsible for much of their "personality". At least not so directly on a specific topic like this.

[operatingthetan]

>Yet more confirmation LLM's have no concept of concepts or context, no intelligence, no self awareness.

The problem is many people seem to believe they have these things and some of those people will put LLMs into situations where this becomes dangerous.

[worldsayshi]

Couldn't this be a flaw in the attention mechanism? Like they need some kind of grounding. An awareness of what they fundamentally should care about and how the thing they are currently giving attention to relates to that?

[Bender]

Words like attention, awareness and care do not apply to computers. At least, not yet. Intelligence and sentience are not applicable to servers. They are just machines with logic states. LLM's are just really cool math formulas with big-data fed into them. Big data is not intelligence. It is a massive data-set sorted, filtered down and interpreted by a language model.

[esprehn]

I assume they meant the Attention process in LLMs, not the human concept of paying attention:

https://en.wikipedia.org/wiki/Attention_(machine_learning)

[slibhb]

LLMs are intelligent by any reasonable standard. Arguing otherwise is like arguing that chess algorithms aren't good at chess when they easily beat the best humans.

[Bender]

I disagree. LLM's are a language model math formulas that interpret and utilize big-data. Take away the math formulas and we are just back to a massive set of data. Adding to that I would suggest not even the purist forms of data meaning that the data-sets include knowledge from the open and anonymous internet and formulaic tuning from the AI owners and operators.

[anon84873628]

Your brain is mostly just a Principal Component Analysis calculator. Take away that "math formula" and you don't have intelligence either.

The LLM weights are not intelligent. But if you give an agent a mutable memory store and allow it to iterate, it is obviously intelligent. Not massively - it's constrained by the context window - but definitely somewhat.

The confusing thing is that their language ability far outpaces their true intelligence, and humans aren't used to that. Normally those things are highly correlated, so it tricks us.

[slibhb]

If you want to talk about whether LLMs are intelligent, you have to define intelligence. "They're just math formulae" isn't a definition of intelligence.

[larodi]

Doesn’t take intelligence to beat a human.

[lukan]

"Like they need some kind of grounding."

A robot body, to really feel the world and get real feedback?

We are working on it. Also on automating the whole production pipeline. Right now a "evil" LLM could indeed not do much, but destroy. But once the whole industry is automate, things are different. I don't believe in AI becoming sentinent and taking over the world any time soon, but I do believe most don't see a danger when it would be inconvenient to see a danger. After all, lots of good and bad sci fi stories about exactly this went into their training.

[chaseadam17]

I'd argue we don't even know what "intelligence" or "self-awareness" mean.

Humans are conscious which means we experience things, then we develop preferences for certain experiences, then we develop skills for achieving those preferences.

Without consciousness, what is there to be aware of? And why would intelligence emerge and/or what end would it serve?

[anon84873628]

Intelligence is the ability to have an internal world model then run simulations on that model to choose an optimal course of action. This is true for humans down to flies. Most of what humans do is still the boring innate stuff; it's just that fancy abstract things like "skydiving" get the most attention.

Clearly other animals have "phenomenological experience" i.e. consciousness / qualia without being as intelligent as humans (or necessarily "self aware"). Many people believe consciousness is simply a side effect of intelligence rather than the other way around.

[mountainriver]

Intelligence is the ability to compress information. World models are just one aspect of that

[mountainriver]

I can’t believe anyone still thinks this given their unbelievable ability to write code.

Self awareness? Probably not. Intelligence? You would have to be high to think that’s not the case.

People are feeling threatened, and rightfully so. LLMs are already insanely intelligent and continue to improve

[yakz]

We shouldn't want them to have self awareness, we shouldn't be seeking to make self-aware actual slaves. We want machines with perception and knowledge, and that are capable of reasoning. But nothing capable of self-determination.

[raffael_de]

Just tried "generate an SVG of a pelican riding a bicycle" for Claude Opus 4.8 Max and of course both legs on same side ... the smartest publicly available model by Anthropic (after Fable) doesn't even successfully simulate understanding the concept of a bicycle.

[mountainriver]

Yet it can write code better than 99% of humans…

It’s just starting to be trained on svgs, which is a really hard problem

[raffael_de]

"99% of humans" is a low bar. Maybe you mean "99% of people who earn money by developing software"?

[WarmWash]

LLMs can't really "see", so I challenge you to draw a pelican on a bike without any visual feedback, just code. Because that is how they are doing it.

Vision tokens for transformers aren't really well solved yet, which is why they can smash a phd math problem and trip over a "count the cats on the chair" problem.

[dinfinity]

> Yet more confirmation LLM's have no concept of concepts or context, no intelligence, no self awareness.

No, it isn't. Look at the absolutely trivial code used to simulate war: https://github.com/kennethpayne01/project_kahn_public/blob/m...

Having LLMs play nonsense toy simulations like this tells us very, very little about whether they would use nukes in real life war.

[doctorpangloss]

If only there was some way you could tell the chatbots what you want them to do...