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

Agents Are Not Enough

arxiv.org

Post reply on HN
Print view View original

121–130 of 165 posts

Re: Agents Are Not Enough

#121

Post by Art9681 »

Terr_ wrote: post #91
lugu wrote: post #87

Earlier quoted context omitted.

I am not sure what you mean by LLM when you say they are professional bullshitter. While it was certainly true for model based on transformers just doing inference, recent models have progressed significantly.

> I am not sure what you mean by LLM when you say they are professional bullshitter. Not parent-poster, but an LLM is a tool for extending a document by choosing whatever statistically-seems-right based on other documents, and it does so with no consideration of worldly facts and no modeling of logical prepositions or contradictions. (Which also relates to math problems.) If it has been fed on documents with logic pu…

I think the problem is the way you are phrasing your argument implies the LLM is always wrong. Consider a simple prompt: "Write a hello world in Python."

Every LLM i've tested gets this correct. In my mind, it can't be both bullshit and correct.

I would argue that the amount of real bullshit returned from an LLM is correlated to the amount of bullshit you give it. Garbage in, garbage out.

In the end, its irrelevant if its a statistical engine or whatever semantics we want to use (glorified autocomplete). If it solved my problem in less time than I perceive I would have solved it without it, bullshit isn't the word I would use to describe the outputs.

In all fairness though, I do get some bullshit responses.

Re: Agents Are Not Enough

#122

Post by daxfohl »

seadan83 wrote: post #119
conception wrote: post #116

Earlier quoted context omitted.

On the other side of the coin, I think people also underestimate the amount of human thinking and intelligence is just completing statistically likely words. Most actions and certainly reactions people do everyday involve very little reasoning. Instead just following the most used neuron.

Human vision works this way. To fix the latency problem (actual event hff happening vs signal transmitted to your brain) human vision is constantly predicting what you should see, your brain tells you that is what you saw (the prediction), and then the brain does reconciliation after the fact. Your brain will scramble for coherency when prediction and reality do not match. This trickery is why it seems like you see e…

True, and this even happens in a brain the size of a flea's.

Which makes one wonder, what is it that makes processing and reconciling millions of visual signals per second "easy", but reasoning through a simple sudoku near impossible?

Re: Agents Are Not Enough

#123

Post by Terr_ »

Art9681 wrote: post #121
Terr_ wrote: post #91

Earlier quoted context omitted.

> I am not sure what you mean by LLM when you say they are professional bullshitter. Not parent-poster, but an LLM is a tool for extending a document by choosing whatever statistically-seems-right based on other documents, and it does so with no consideration of worldly facts and no modeling of logical prepositions or contradictions. (Which also relates to math problems.) If it has been fed on documents with logic pu…

I think the problem is the way you are phrasing your argument implies the LLM is always wrong. Consider a simple prompt: "Write a hello world in Python." Every LLM i've tested gets this correct. In my mind, it can't be both bullshit and correct. I would argue that the amount of real bullshit returned from an LLM is correlated to the amount of bullshit you give it. Garbage in, garbage out. In the end, its irrelevant i…

> In my mind, it can't be both bullshit and correct.

It's easy for bullshitters to say some true things, but it doesn't change the nature of the process that got the results. Ex:

________

Person A: "The ghost of my dead gerbil whispers unto me the secrets of the universe, and I am hearing that the local volcano will not erupt today."

Person B: "Bullshit."

[24 hours later]

Person A: "See? I was correct! I demand an apology for your unjustified comment."

Re: Agents Are Not Enough

#125

Post by stevenwoo »

daxfohl wrote: post #122
seadan83 wrote: post #119

Earlier quoted context omitted.

Human vision works this way. To fix the latency problem (actual event hff happening vs signal transmitted to your brain) human vision is constantly predicting what you should see, your brain tells you that is what you saw (the prediction), and then the brain does reconciliation after the fact. Your brain will scramble for coherency when prediction and reality do not match. This trickery is why it seems like you see e…

True, and this even happens in a brain the size of a flea's. Which makes one wonder, what is it that makes processing and reconciling millions of visual signals per second "easy", but reasoning through a simple sudoku near impossible?

Are you sure about fleas? I thought we know this does not happen with frogs as far as we can tell https://courses.csail.mit.edu/6.803/pdf/lettvin.pdf

Re: Agents Are Not Enough

#126

Post by bee_rider »

jerojero wrote: post #118
Upvoter33 wrote: post #99

Earlier quoted context omitted.

"If AGI ... is possible" I don't get this line of thinking. AGI already exists - it's in our heads! So then the question is: is what's in our heads magic, or can we build it? If you think it's magic, fine - no point arguing. But if not, we will build it one day.

The brain is such an intractable web of connections that it has been really difficult to properly make sense of it. We can't really talk too much about the differences between the intelligence of a dog and the intelligence of a human; in real terms. It seems as though humans might have more connections, different types of cells but then again; there's species out there that also have types of neurons we don't have an…

What processes in our cells make use of quantum mechanics? (I mean in some sense everything is quantum mechanics, but cells are quite big in a quantum mechanics sense. I’d imagine they are mostly classical).

Re: Agents Are Not Enough

#127

Post by notamy »

bee_rider wrote: post #126
jerojero wrote: post #118

Earlier quoted context omitted.

The brain is such an intractable web of connections that it has been really difficult to properly make sense of it. We can't really talk too much about the differences between the intelligence of a dog and the intelligence of a human; in real terms. It seems as though humans might have more connections, different types of cells but then again; there's species out there that also have types of neurons we don't have an…

What processes in our cells make use of quantum mechanics? (I mean in some sense everything is quantum mechanics, but cells are quite big in a quantum mechanics sense. I’d imagine they are mostly classical).

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

> Mitochondria have been demonstrated to utilize quantum tunneling in their function as the powerhouse of eukaryotic cells.

Re: Agents Are Not Enough

#128

Post by energy123 »

daxfohl wrote: post #115
energy123 wrote: post #103

Earlier quoted context omitted.

You could, but in many cases you wouldn't want to. You will get superior results with a fixed compute budget by relying on external tool use (where "tool" is defined liberally, and can include smaller narrow neural nets like GraphCast & AlphaGo) rather that stuffing all tools into a monolithic model.

Isn't that what the original resnet project disproved? Rather than trying to hand-manicure what the NN should look for, just make it deep enough and give it enough training data, and it'll figure things out on its own, even better than if we told it what to look out for. Of course, cost-wise and training time wise, we're probably a long way off from being able to replicate that in a general purpose NN. But in theory,…

I'm not proposing hand-engineering anything, though. I'm proposing giving the AI tools, like a calculator API, a code interpreter, search, and perhaps a suite of narrow AIs that are superhuman in niche domains. The AI with tool use should outperform a competitor AI that doesn't have access to these tools, all else equal. The reason should be intuitive: the AI with tool use can dedicate more of its compute to the reasoning that is not addressed by the available tools. I don't think my views here are inconsistent with The Bitter Lesson.

Re: Agents Are Not Enough

#129

Post by rcarmo »

georgestrakhov wrote: post #28

IMHO, the word agent is quickly becoming meaningless. The amount of agency that sits with the program vs. the user is something that changes gradually. So we should think about these things in terms of how much agency are we willing to give away in each case and for what gain[1]. Then the ecosystem question that the paper is trying to solve will actually solve itself, because it is already the case today that in many…

I think people keep conflating agency with agents, and that they are actually two entirely different things in real life. Right now agents have no agency - they do dot independently come up with new approaches, they’re mostly task-oriented.
Post reply on HN