Ilya Sutskever NeurIPS talk [video]

https://www.youtube.com/watch?v=1yvBqasHLZs

by mfiguiere • 2 years ago

309 239 2 years ago

skissane3• 2 years ago

“Pre-training as we know it will unquestionably end,” Sutskever said onstage.
“We’ve achieved peak data and there’ll be no more.”
During his NeurIPS talk, Sutskever said that, while he believes existing data can still take AI development farther, the industry is tapping out on new data to train on. This dynamic will, he said, eventually force a shift away from the way models are trained today. He compared the situation to fossil fuels: just as oil is a finite resource, the internet contains a finite amount of human-generated content.
“We’ve achieved peak data and there’ll be no more,” according to Sutskever. “We have to deal with the data that we have. There’s only one internet.”

What will replace Internet data for training? Curated synthetic datasets?

There are massive proprietary datasets out there which people avoid using for training due to copyright concerns. But if you actually own one of those datasets, that resolves a lot of the legal issues with training on it.

For example, Getty has a massive image library. Training on it would risk Getty suing you. But what if Getty decides to use it to train their own AI? Similarly, what if News Corp decides to train an AI using its publishing assets (Wall Street Journal, HarperCollins, etc)?

_aavaa_• 2 years ago

just as oil is a finite resource, the internet contains a finite amount of human-generated content.

I guess now they’re being explicit about the blatantly extractive nature of these businesses and their models.

kibae• 2 years ago

I always suspected that bots on Reddit were used to gain karma and then eventually sell the account, but maybe they're also being used for some kind of RLHF.

vitorgrs• 2 years ago

Not sure if this was a good example. Getty already license their images to Nvidia.

And they already have a generative image service... I believe it's power by Nvidia model.

popularonion• 2 years ago

What will replace Internet data for training? Curated synthetic datasets?

Enter Neuralink

zxexz3• 2 years ago

Really not sure what you mean by this, could you explain?

YetAnotherNick2• 2 years ago

Humans doesn't need trillions of tokens to reason or ability to know what they know. While a certain part of it comes from evolution, I think we have already matched the part that came from evolution using internet data, like basic language skills, basic world modelling. Current pretraining takes lot more data than a human would, and you don't need to look into all Getty images to draw a picture and so would a self aware/improving model(whatever that means).

To reach expert level in any field, just training next tokens for internet data or any data is not the solution.

exe343• 2 years ago

Humans doesn't need trillions of tokens

I wonder about that. we can fine tune on calculus with much fewer tokens, but I'd be interested in some calculations of how many tokens evolution provides us (it's not about the DNA itself, but all the other things that were explored and discarded and are now out of reach) - but also the sheer amount of physics learnt by a baby by crawling around and putting everything in its mouth.

bronco21016• 2 years ago

Humans doesn't need trillions of tokens to reason or ability to know what they know.

It seems to me by the time we’re 5-6 we’ve likely already been exposed to trillions of tokens. Just think of how many hours of video and audio tokens have already come to your brain by that point. We also have constant input from senses like touch and proprioception that help shape our understanding of the world around us.

I think there are plenty more tokens freely available out in the world. We just haven’t figured out how to have machines capture them yet.

menaerus9• 2 years ago

What will replace Internet data for training? Curated synthetic datasets?

Perhaps a different take at this could be: if I wanted to train "state law" LLM that is exceedingly good in interpreting state law, what are the obstacles to download all the law and regulations material for given state that will allow me to train LLM such that it becomes 95th percentile of all law trainees and lawyers.

In that case, and my point is, that we already don't need an "Internet". We just need a sufficiently sized and curated domain-specific dataset and the result we can get is already scary. "State law" LLM was just an example but the same logic applies to basically any other domain - want a domain-specific (LLM) expert? Train it.

pas2• 2 years ago

you need context for the dry statutes

sure, you download all the legal arguments, and hope that putting all this on top of a general LLM which has enough context to deal with usual human, American, contemporary stuff

the argument is that it's not really enough for the next jump (as it would need "exponentially" more data) as far as a I understand

losvedir• 2 years ago

That's kind of going in a different direction. The big picture is that LLMs have until this point gotten better and better from larger datasets alone. See "The Bitter Lesson". But now we're running out of datasets and so the only way we know of to improve models' reasoning abilities and everything, is coming to an end.

You're talking about fine tuning, which yes is a technique that's being used and explored in different domains, but my understanding is it's not a very good way for models to acquire knowledge. Instead larger context windows and RAG works better for something like case law. Fine tuning works for things like giving models a certain "voice" in how they produce text, and general alignment things.

At least that's my understanding as an interested but not totally involved follower of this stuff.

yeahwhatever10• 2 years ago

The problem remains the size of the dataset. You aren't going to get large enough datasets in these specific domains.

sharih2• 2 years ago

the big frontier models already have all laws, regulations and cases memorized/trained on given they are public. the real advancement is in experts codifying their expertise/reasoning for models to learn from. legal is no different from other fields in this.

fidotron• 2 years ago

What will replace Internet data for training? Curated synthetic datasets?

My take is that the access Meta, Google etc. have to extra data has reduced the amount of research into using synthetic data because they have had such a surplus of it relative to everyone else.

For example, when I've done training of object detectors (quite out of date now) I used Blender 3D models, scripts to adjust parameters, and existing ML models to infer camera calibration and overlay orientation. This works amazingly well for subsequently identifying the real analogue of the object, and I know of people doing vehicle training in similar ways using game engines.

There were several surprising tactical details to all this which push the accuracy up dramatically and you don't see too widely discussed, like ensuring that things which are not relevant are properly randomized in the training set, such as the surface texture of the 3D models. (i.e. putting random fractal patterns on the object for training improves how robust the object detector is to disturbance in reality).

robg• 2 years ago

The ones that stand out to me are industries like pharmaceuticals and energy exploration, where the data silos are the point of their (assumed) competitive advantages. Why even the playing by opening up those datasets when keeping them closed locks in potential discoveries? Open data is the basis of the Internet. But whole industries are based on keeping discoveries closely guarded for decades.

seydor• 2 years ago

Robots can acquire data on their own (hopefully not via human dissection)

rapjr9• 2 years ago

This is indeed what I thought he was saying, AI needs to dynamically learn, just training on static data sets is no longer enough to advance. So continuous learning is the future, and the best source of data for continuous learning is people themselves. Don't know what form that might take, instrumenting lots of people with sensors? Robots interacting with people? Self driving cars learning from human drivers? (already happening) Ingesting all video from surveillance cameras? Whatever form the input data takes, continuous learning would be an advance in high level AI. There's been work on it over the decades, not sure how that work might relate to recent LLM work.

parkaboy• 2 years ago

I wonder if we will see (or already are/have been seeing) the XR/smart glasses space heat up. Seems eventually like a great way to generate and hoover up massive amounts of fresh training data.

RicoElectrico• 2 years ago

I think we're not close to running out of training data. It's just that we would like knowledge, but not necessary behavior of said texts. LLMs are very bad at recalling popular memes (known by any seasoned netizen) if they had no press coverage. Maybe training with 4chan isn't as pointless if you could make it memorize it, but not imitate it.

Also, what about movie scripts and song lyrics? Transcripts of well known YouTube videos? Hell, television programs even.

stavros• 2 years ago

We've run out of training data that definitely did not contain LLM outputs.

gcollard-• 2 years ago

All the publicly accessible sources you mentioned have already been scraped or licensed to avoid legal issues. This is why it’s often said, “there’s no public data left to train on.”

For evidence of this, consider observing non-English-speaking young children (ages 2–6) using ChatGPT’s voice mode. The multimodal model frequently interprets a significant portion of their speech as “thank you for watching my video,” reflecting child-like patterns learned from YouTube videos.

zozbot2343• 2 years ago

Synthetic datasets are useless (other than for very specific purposes, such as enforcing known strong priors, and even then it's way better to do it directly by changing the architecture). You're better off spending that compute by making multiple passes over the data you do have.

HeatrayEnjoyer3• 2 years ago

This is contrary to what the big AI labs have found. Synthetic data is the new game in town.

toxik• 2 years ago

Most priors are not encodable as architecture though, or only partially.

oldgradstudent• 2 years ago

There are massive proprietary datasets out there which people avoid using for training due to copyright concerns.

The main legal concern is their unwillingness to pay to access these datasets.

zozbot2343• 2 years ago

Yup, there's also a huge amount of copyright-free, public domain content on the Internet which just has to be transcribed, and would provide plenty of valuable training to a LLM on all sorts of varied language use. (Then you could use RAG over some trusted set of data to provide the bare "facts" that the LLM is supposed to be talking about.) But guess what, writing down that content accurately from scans costs money (and no, existing OCR is nowhere near good enough), so the job is left to purely volunteer efforts.

numpad0• 2 years ago

What will replace Internet data for training?

It means unlimited scaling with Transformer LLM is over. They need a new architecture that scales better. Internet data respawns when they click [New Game...], oil analogy is an analogy and not a fact, but anyways total amount available in a single game is finite so combustion efficiency matters.

sureglymop• 2 years ago

Perhaps these models will be more personalized and there will be more personal data collection.

I am currently building a platform for heavy personal data collection including a keylogger, logging mouse positions and window focus, screenshots à la recall, open browser tabs and much more. My goal is to gather data now that may become useful later. It's mainly for personal use but I'd he surprised if e.g. iphones weren't headed in the same direction.

lo_fye• 2 years ago

I think this will be the one thing that causes Google to revive its plan to scan all books in existence. They had started it, and built the machines to do it, and were making good progress... until Copyright hit them. BUT if they're not making the full text publicly accessible, and are "only" training AI on it, who knows if that would still be a problem. It's definitely a vast treasure trove of information, often with citations, and (presumably) hyper-linkable sources.

neom• 2 years ago

Full talk is interesting: https://www.youtube.com/watch?v=YD-9NG1Ke5Y

CuriousSkeptic• 2 years ago

On the slide of body/brain weight relation he highlighted the humanids difference in scaling

What he didn't mention, that I found interesting, was that the same slide also highlighted a hard ceiling for non-humanids at the same point

imranhou• 2 years ago

This is a very interesting point, in some ways the implicit belief is that we just need to get beyond the 700g limitation in terms of scaling LLM models and we would get human intelligence/superintelligence. I admit I didn't really get the body/brain analogy, I would have been better satisfied with a simpler graph of brain weight to intelligence with a scaling barrier of 700g.

stretchwithme• 2 years ago

AIs will need to start asking people questions. Should make for some very strange phone calls.

wslh2• 2 years ago

That's a good point. I think most people use LLMs by asking questions and receiving answers. But if you reverse the dynamic and have the LLM interview you instead, where you simply respond to its questions, you'll notice something interesting: the LLM as an interviewer is far less "smart" than it is when simply providing answers. I’ve tried it myself, and the interview felt more like interacting with ELIZA [1].

There seemed to be a lack of intent when the LLM was the one asking the questions. This creates a reverse dynamic, where you become the one being "prompted" and this dynamic could be worth studying or adjusting further

[1] https://en.wikipedia.org/wiki/ELIZA

airstrike• 2 years ago

Which LLM did you perform that test with?

Barrin92• 2 years ago

There seemed to be a lack of intent when the LLM was the one asking the questions

There doesn't just seem to be lack of intent, there is no intent, because by the nature of its architecture these systems are just set of weights with a python script attached to them asking you to give you one more token over and over.

There's no needs, drives, motivations, desires or any other part of the cognitive architecture of humans in there that produce genuine intent.

staticman22• 2 years ago

Did you ask the AI to roleplay? All it does is predict what text comes next. Telling it it is role-playing as a perceptive journalist or prominent psychologist should change its predictions.

In some cases using it where you have API access to the system prompt will allow a greater difference in behavior.

zxexz3• 2 years ago

I can't help but feel that this talk was a lot of...fluff?

The synopsis, as far as my tired brain can remember:

- Here's a brief summary of the last 10 years

- We're reaching the limit of our scaling laws, because we've trained on all the data we have available on the limit

- Some things that may be next are "agents", "synthetic data", and improving compute

- Some "ANNs are like biological NNs" rehash that would feel questionable if there was a thesis (which there wasn't? something about how body mass vs. brain mass are positively correlated?)

- 3 questions, the first was something about "hallucinations" and whether a model be able to understand if it is hallucinating? Then something that involved cryptocurrencies, and then a _slightly_ interesting question about multi-hop reasoning

coeneedell• 2 years ago

I attended this talk in person and some context is needed. He was invited for the “test of time” talk series. This explains the historical part of the talk. I think his general persona and association with ai led to the fluffy speculation at the end.

I notice with Ilya he wants to talk about these out there speculative topics but defends himself with statements like “I’m not saying when or how just that it will happen” which makes his arguments impossible to address. Stuff like this openly invites the crazies to to interact with him, as seen with the cryptocurrency question at the end.

Right before this was a talk reviewing the impact of GANs that stayed on topic for the conference session throughout.

mrbungie5• 2 years ago

I mean he repeateadly gave some hints (even if just for the lulz and not seriously) that the audience is at least partially composed of people with little technical background or AI bros. An example is when he mentioned LSTMs and said "many of you may never seen before". Even if he didn't mean it, ironically it ended being spot on when the crypto question came.

ITjournalist• 2 years ago

"a talk reviewing the impact of GANs"

Is that available online?

killerstorm10• 2 years ago

Well, it looks like the entire point was "you can no longer expect a capability gain from a model with a bigger ndim trained on a bigger internet dump".

That's just one sentence, but it's pretty important. And while many people already know this, it's important to hear Sutskever say this. So people know it's a common knowledge.

The rest is basically intro/outro.

mrbungie5• 2 years ago

It is very important to have at least some kind of counterweight vs OpenAI/sama predicting AGI for 2025/2026.

lottin2• 2 years ago

He makes a good point. But then he jumps to "models will be self-aware". I fail to see any logical connection.

KuriousCat• 2 years ago

So, are we headed towards a bitter-sweet zone and modeling is going to get more prominent once again? massive datasets going to take a backseat?

bflesch4• 2 years ago

For me the questions were a big red flag. Fluff questions about crypto, human rights for AI and then "autocorrect" for AI. Obviously people who ask questions at conference talks are a special type of person but these topics scream to me that there's so many grifters in the AI space right now it might as well drown authentic research.

By now, most of the fundamental contributors are multi-millionaires with cushy contracts. Various labs & departments have their big fat funding for AI research topics. They will be able to spend next 10 years on synthetic data, or "agents", or ensuring that no breasts are in auto-generated images; but somehow it doesn't feel to me like there'll be a lot of fundamental progress.

/remindme 10 years

throwaway20372• 2 years ago

Thank you to summarise the video. No trolling here: I am surprised that no one asked an LLM to summarise the video, then post the result here as a comment (with LLM "warning" of course).

benchmarkist6• 2 years ago

Here you go: https://g.co/gemini/share/852449bad133

Havoc• 2 years ago

fluff

I’d imagine even for the best of minds it’s hard to always come up with something profound on demand

sigmar• 2 years ago

I found this week's DeepMind podcast with Oriole Vinyals to be on similar topics as this talk (current situation of LLMs, path ahead with training) but much more interesting: https://pca.st/episode/0f68afd5-2b2b-4ce9-964f-38193b7e8dd3

sega_sai• 2 years ago

Very thought provoking. One of the things was not clear to me, what does he mean by 'agentic' intelligence?

kgeist• 2 years ago

When it autonomously performs tasks on behalf of the user, without their intervention. It sets goals, plans actions, etc. by itself.

ototot• 2 years ago

Example of 'agentic': https://blog.google/technology/google-deepmind/google-gemini...

ethbr12• 2 years ago

It's the current industry buzzword for action-oriented LLMs.

Instead of generating text, tightening/formalizing the loop around planning, executing, analyzing results, and replanning.

As far as buzzwords go, it's far from the worst, as it captures the essentials -- creating semi-autonomous agents.

ed• 2 years ago

Based on the context Ilya is not referring to that kind of agent. He’s referring to something more fundamental (which I was curious about, too).

legel• 2 years ago

I’m glad Ilya starts the talk with a photo of Quoc Le, who was the lead author of a 2012 paper on scaling neural nets that inspired me to go into deep learning at the time.

His comments are relatively humble and based on public prior work, but it’s clear he’s working on big things today and also has a big imagination.

I’ll also just say that at this point “the cat is out of the bag”, and probably it will be a new generation of leaders — let us all hope they are as humanitarian — who drive the future of AI.

chipsrafferty• 2 years ago

Literally a zero chance that the new generation of leaders of artifical intelligence will be humanitarian.

mrbungie5• 2 years ago

Let us all hope that they will be as humanitarian as they can be, but let's not forget they are still just human beings.

tikkun• 2 years ago

As context on Ilya's predictions given in this talk, he predicted these in July 2017:

Within the next three years, robotics should be completely solved [wrong, unsolved 7 years later], AI should solve a long-standing unproven theorem [wrong, unsolved 7 years later], programming competitions should be won consistently by AIs [wrong, not true 7 years later, seems close though], and there should be convincing chatbots (though no one should pass the Turing test) [correct, GPT-3 was released by then, and I think with a good prompt it was a convincing chatbot]. In as little as four years, each overnight experiment will feasibly use so much compute capacity that there’s an actual chance of waking up to AGI [didn't happen], given the right algorithm — and figuring out the algorithm will actually happen within 2–4 further years of experimenting with this compute in a competitive multiagent simulation [didn't happen].

Being exceptionally smart in one field doesn't make you exceptionally smart at making predictions about that field. Like AI models, human intelligence often doesn't generalize very well.

padolsey• 2 years ago

exceptionally smart at making predictions

Is anyone though? Genuine question. I don't have much faith in predictions anymore.

qeternity• 2 years ago

No, very few for things with this much uncertainty.

Most of it is survivorship bias: if you have a million people all making predictions with coin flip accuracy, somebody is going to get a seemingly improbable number correct.

exe343• 2 years ago

so your prediction is that most predictions will be wrong?

ethbr12• 2 years ago

Predictions predicated on technological advancement are tricky: there's a reason breakthroughs are called breakthroughs.

_giorgio_2• 2 years ago

he just wanted money from investors, that's why he used such short limits

https://openai.com/index/elon-musk-wanted-an-openai-for-prof...

2/3/4 will ultimately require large amounts of capital. If we can secure the funding, we have a real chance at setting the initial conditions under which AGI is born.

InkCanon• 2 years ago

For all the discussion about it, this is the simple answer. It's not an engineering or scientific prediction, it's a line from a pitch deck.

sangnoir• 2 years ago

It is important to note the context: he it was in a private email to an investor with vested interests in those fields, and someone who is also prone to giving over-optimistic timelines ("Tobo-taxis will be here next year, for sure" since 2015)

_giorgio_2• 2 years ago

What a stupid talk.

They gave 15 minutes to one of the most competent scientist.

A joke.

ldenoue• 2 years ago

LLM corrected transcript (using Gemini Flash 8B over the raw YouTube transcript) https://www.appblit.com/scribe?v=YD-9NG1Ke5Y#0

oezi• 2 years ago

How do you prevent Gemini from just swallowing text after some time?

Audio transcript correction is one area where I struggle to see good results from any LLM unless I chunk it to no more than one or two pages.

Or did you use any tool?

belter2• 2 years ago

It’s surprising that some prominent ML practitioners still liken transformer ‘neurons’ to actual biological neurons...

Real neurons rely on spiking, ion gradients, complex dendritic trees, and synaptic plasticity governed by intricate biochemical processes. None of which apply to the simple, differentiable linear layers and pointwise nonlinearities in transformers.

Are there any reputable neuroscientists or biologists endorsing such comparisons, or is this analogy strictly a convention maintained by the ML community? :-)

martindbp2• 2 years ago

You have to remember what came before 2012: SVMs, Random Forests etc, absolutely nothing like the brain (yes, NNs are old, but 2012 was the start of the deep learning revolution). With this frame of reference, the brain and neural networks are both a kind of Connectionism with similar properties, and I think it makes perfect sense to liken them with each other, draw inspiration from one and apply it to the other.

signa11• 2 years ago

sorry, but i think neural-networks came way before 2012, notably the works of rumelhart, mccleland etc. see the 2 volume "parallel distributed processing" to read almost all about it.

the book(s): https://direct.mit.edu/books/monograph/4424/Parallel-Distrib...

a-talk: https://www.youtube.com/watch?v=yQbJNEhgYUw

zitterbewegung• 2 years ago

Neural Networks are 200 years old (Legendre and Gauss defined Feed forward neural networks). Deep learning. The real difference between traditional ones and deep learning is a hierarchy of layers (hidden layers) which do different things to accomplish a goal. Even the concept of training is to provide weights on the neural network and there are many algorithms to do refinement, optimization and the network design.

belter2• 2 years ago

It is also odd to see such a weak argument as the brain-to-body mass ratio being used, as here: https://youtu.be/YD-9NG1Ke5Y?t=593

If this metric were truly indicative, what should we make of the remarkable ratios found in small birds (1:12), tree shrews (1:10), or even small ants (1:7)?

https://en.wikipedia.org/wiki/Brain%E2%80%93body_mass_ratio

zk4x• 2 years ago

What came before was regression. Which is to this day no 1 method if we want something interpretable, especially if we know which functions our variables follow. And self attention is very similar to correlation matrix. In a way neural networks are just bunch of regression models stacked on top of each other with some normalization and nonlinearity between them. It's cool however how closely it resembles biology.

bflesch4• 2 years ago

Sorry, but they just called it "neuron" to sound nicer.

criddell• 2 years ago

Is that wildly different from me calling a data structure where a parent node has child nodes a tree?

wrs• 2 years ago

Depends — do you then start claiming that because your data structure is like a tree, it’s surely going to start bearing fruit and emitting oxygen?

modzu• 2 years ago

what color are neurons? is that relevant? ml has proven that artificial networks can think. the other stuff may be necessary to do other things, or maybe simply evolved to support the requisite biological structures. ml is of course inspired by biology, but that does not mean we need to simulate everything.

bflesch4• 2 years ago

ml has proven that artificial networks can think

nope, at this point it's an ad-free version of google that summarizes results without linking to source websites.

chpatrick• 2 years ago

You don't need to simulate every atom in a planet to predict its orbit. A mathematical neuron could have similar function to a real one even if it works completely differently.

sourcepluck• 2 years ago

Reading the replies to your comment, I think maybe the answer to your simple question is: "no". I also wonder if any "serious comparisons" have been made, and would be interested to read about it! A good question, I think.

syassami• 2 years ago

https://www.bloomberg.com/news/articles/2024-12-13/liquid-ai...

curious_cat_163• 2 years ago

Not excusing the lack of caveat in his talk but IMO, the old adage of: "All models are wrong, but some are useful." applies here.

sensanaty• 2 years ago

just as oil is a finite resource, the internet contains a finite amount of human-generated content.

The oil comparison is really apt. Indeed, let's boil a few more lakes dry so that Mr Worldcoin and his ilk can get another 3 cents added to their net worth, totally worth it.

seizethecheese• 2 years ago

I understand the oil analogy, but not your leap. What lake is getting boiled?

lennxa• 2 years ago

they are referring to the "water footprint" of LLMs. https://deepgram.com/learn/how-ai-consumes-water

olddog2• 2 years ago

So much knowledge in the world is locked away with empiric experimentation being the only way to unlock it, and compute can only really help that experimentation become more efficient. Something still has to run a randomized controlled trial on an intervention and that takes real time and real atoms to do.

killthebuddha4• 2 years ago

One thing he said I think was a profound understatement, and that's that "more reasoning is more unpredictable". I think we should be thinking about reasoning as in some sense exactly the same thing as unpredictability. Or, more specifically, useful reasoning is by definition unpredictable. This framing is important when it comes to, e.g., alignment.

bondarchuk2• 2 years ago

Not necessarily true when you think about e.g. finding vs. verifying a solution (in terms of time complexity).

killthebuddha4• 2 years ago

IMO verifying a solution is a great example of how reasoning is unpredictable. To say "I need to verify this solution" is to say "I do not know whether the solution is correct or not" or "I cannot predict whether the solution is correct or not without reasoning about it first".

narrator• 2 years ago

Reasoning by analogy is more predictable because it is by definition more derivative of existing ideas. Reasoning from first principles though can create whole new intellectual worlds by replacing the underpinnings of ideas such that they grow in completely new directions.

mike_hearn• 2 years ago

Wouldn't it be the reverse? The word unreasonable is often used as a synonym for volatile, unpredictable, even dangerous. That's because "reason" is viewed as highly predictable. Two people who rationally reason from the same set of known facts would be expected to arrive at similar conclusions.

I think what Ilya is trying to get at here is more like: someone very smart can seem "unpredictable" to someone who is not smart, because the latter can't easily reason at the same speed or quality as the former. It's not that reason itself is unpredictable, it's that if you can reason quickly enough you might reach conclusions nobody saw coming in advance, even if they make sense.

bflesch4• 2 years ago

they only arrive at the same conclusion if they both have the same goal.

one could be about maximising wealth while respecting other human beings, the other could be about maximising wealth without respecting other human beings.

Both could be presented same facts and 100% logical but arrive at different conclusions.

killthebuddha4• 2 years ago

Your second paragraph is basically what I'm saying but with the extension that we only actually care about reasoning when we're in these kinds of asymmetric situations. But the asymmetry isn't about the other reasoner, it's about the problem. By definition we only have to reason through something if we can't predict (don't know) the answer.

I think it's important for us to all understand that if we build a machine to do valuable reasoning, we cannot know a priori what it will tell us or what it will do.

liuliu• 2 years ago

I think many of replies here to you missing is the word he uses is "unpredictable". It is not "surprising", "unverifiable" or "unreasonable".

"Prediction" associated in this particular talk is about "intuition": what human can do in 0.1 second. And a most powerful reasoning model by its definition will arrive at "unintuitive" answer because if it is intuitive, it will arrive at the same answer much sooner without long chain of "reasoning". (I also want to make distinction "reasoning" here is not the same as "proof" in mathematical sense. In mathematics, an intuitive conclusion can require extrodinary proof.)

bmitc• 2 years ago

Are you sure that's what he was referring to? In other words, you don't think he was meaning that getting more reasoning out of models is an unpredictable process and not saying that reasoning is unpredictable.

billyzs• 2 years ago

To me the chess AI example he used was perhaps not the most apt. Human players may not be able to reason on as far a horizon as AI and therefore find some of AI's moves perplexing, but they can be more or less sure that a Chess AI is optimizing for the same goal under the same set of rules with them. With Reasoners, alignment is not given. They may be reasoning under an entirely different set of rules and cost functions. On more open ended questions, when Reasoners produce something that human don't understand, we can't easily say whether it's a stroke of genius, or misaligned thoughts.

linsomniac• 2 years ago

ISTR reading back in the mid '90s, in a book on computing history which I have long since forgotten the exact name/author of, something along the lines of:

In the mid '80s it was largely believed among AI researchers that AI was largely solved, it just needed computing horsepower to grow. Because of this AI research stalled for a decade or more.

Considering the horsepower we are throwing at LLMs, I think there was something to at least part of that.

LampCharger• 2 years ago

Ha. Do people understand time for humanity to save itself is running out. What is the point of having a super human AGI if there's no human civilization for which it can help?

HeatrayEnjoyer3• 2 years ago

"We can totally control an entity with 10^x faster and stronger intelligence than us. There is no way this could go wrong, in fact we should spend all of our money building it as soon as possible."

talldayo• 2 years ago

We can totally control an entity with 10^x faster and stronger intelligence than us.

Unless you're referencing an unreleased model that can count the number of 'r' occurrences in "strawberry" then I don't even think we're dealing with .01*10^x intelligence right now. Maybe not even .001e depending on how bad of a Chomsky apologist you are.

ilaksh• 2 years ago

Larger models are more robust reasoners. Is there a limit? What if you make a 5 TB model trained on a lot of multimodal data where the language information was fully grounded in videos and images etc. Could more robust reasoning be that simple?

ryoshu• 2 years ago

It could be simpler. Humans don't need 5TB of data to reason.

Workaccount2• 2 years ago

The amount of data needed to train a human brain is enormous.

Philpax• 2 years ago

Think about the sheer amount of data that you receive through your five senses over 570,000,000 seconds/18 years. It's a lot, lot more than 5TB.

hackandthink• 2 years ago

What kind of reasing is he talking about?

Why should it be unpredictable?

  Deductive Reasoning
  Inductive Reasoning
  Abductive Reasoning
  Analogical Reasoning
  Pragmatic Reasoning
  Moral Reasoning
  Causal Reasoning
  Counterfactual Reasoning
  Heuristic Reasoning
  Bayesian Reasoning

(List generated by ChatGPT)

swyx2• 2 years ago

this is stolen and reposted content. the source video is here. https://youtu.be/1yvBqasHLZs?si=pQihchmQG3xoeCPZ

dang• 2 years ago

Ok, we've changed to that from https://www.youtube.com/watch?v=YD-9NG1Ke5Y. Thanks!

error93482• 2 years ago

It would be great if all NeurIPS talks were accessible for free like this one. I understand they generate some revenue from online ticket sales, but it would be a great resource. Maybe some big org could sponsor it.

swyx2• 2 years ago

they are - on a 1 month delay. https://slideslive.com/neurips-2023 last year.

so if you have the patience, wait, if no patience, pay. fair?

we did this for ai.engineer too except we believe in youtube a bit more for accessibility/discoverability. https://www.youtube.com/@aiDotEngineer/videos

error93482• 2 years ago

Wow, thanks for the correction. Didn't know this existed - to be fair, I only found a preview last year I tried, and paid up.

davidmurphy• 2 years ago

sweet! thanks for posting

IWeldMelons• 2 years ago

How about NeurVA, NeurTN or NeurOLED?

anticensor• 2 years ago

IPS in NeurIPS stands for Information Processing Systems.

niyyou2• 2 years ago

I’ll take the risk of hurting the groupies here. But I have a genuine question: what did you learn from this talk? Like… really… what was new? or potentially useful? or insightful perhaps? I really don’t want to sound bad-mouthed but I‘m sick of these prophetic talks (in this case, the tone was literally prophetic—with sudden high and grandiose pitches—and the content typically religious, full of beliefs and empty statements.

niyyou2• 2 years ago

Precision: « pre-training data is exhausted » everyone has been saying that for a while now. The graph plotting body mass against brain mass… what does it say exactly? (where is the link to the prior point on data?). I think we would all benefit from being more critical here and stop idealizing these figures. I believe they have no more clue that any other average ML researcher on all these questions.

XenophileJKO• 2 years ago

The other thing that bugged me is the built in assumption that today's model have learned everything there is to learn from the Internet corpus. This is quite easy to disprove. Both in factual retention, but also meta cognition on the context of the content.

cma• 2 years ago

Also left out the Baidu scaling laws paper from 2017, and his circle has a history of a kind of citation ring type thing leaving earlier stuff out

https://research.baidu.com/Blog/index-view?id=89

random3• 2 years ago

What everyone could learn is to check their (and their communities') assumptions from not long ago. Who saw this, who didn't. Based on this many can confirm their beliefs and others can realize they're clueless. In either case, there's something to be learned but more to be learned when you realize you were wrong.

fullstackwife• 2 years ago

Today I searched for early discussions about Transformer here on HN, and my observation is that back in 2019 nobody in HN comments predicted what is going to happen. It was a niche topic most of the commenters ignored, no strong opinions.

Probably what we are discussing here is not the next breakthrough...

29athrowaway3• 2 years ago

From your reaction I guess you were expecting a talk about a NeurIPS 2024 paper.

This is a different situation. There's the "NeurIPS 2024 Test of Time Paper Awards" where they award a historical paper. In this case, a paper from 2014 was awarded and his talk is about that and why it passed the test of time.

https://blog.neurips.cc/2024/11/27/announcing-the-neurips-20...

The title chosen for the HN submission leaves out that important context. So that's why you are disappointed now.

p1esk• 2 years ago

I am also disappointed, and I have not missed the context. The talk is empty for anyone who follows the field for more than two years, and especially for those who are familiar with his 2014 paper. Yes, he had amazing insight and intuition behind modern LLM breakthroughs, and yes, he probably earned the right to sound "prophetic", but he could have provided some interesting personal anecdotes about how the paper was written, or some fresh ideas in "What Comes Next" section of his presentation.

abetusk2• 2 years ago

I'll give my take:

* Before the current renaissance of neural networks (pre ~2014ish), it was unclear that scaling would work. That is, simple algorithms on lots of data. The last decade has pretty much addressed that critique and it's clear that scaling does work to a large extent, and spectacularly so.

* Much of the current neural network models and research are geared towards "one-shot" algorithms, doing pattern matching and giving an immediate result. Contrast this with search which needs to do inference time compute or search.

* The exponential increase in power means that neural network models are quickly sponging up as much data as they can find and we're quickly running into the limits of science, art and other data that humans have created in the last 5k years or so.

* Sutskever points out, as an analogy, nature has created a better model for humans (the brain to mass ratio for animals) with hominids finding more efficient compute than other animals, even ones with much larger brains and neuron count.

* Sutskever is advocating for better models, presumably focusing on inference time computer more.

In some sense, we're coming a bit full circle where people who were advocating for pure scaling (simple algorithms + lots of data) for learning are now advocating for better algorithms, presumably with a focus on inference time compute (read: search).

I agree that it's a little opaque, especially for people who haven't been paying attention to past and current research, but this message seems pretty clear to me.

Noam Brown had a talk recently titled "Parables on the Power of Planning in AI" [0] which addresses this point more head on.

I will also point out that the scaling hypothesis is closely related to "The Bitter Lesson" by Rich Sutton [1]. Most people focus on the "learning" aspect of scaling but "The Bitter Lesson" very clearly articulates learning and search as the methods most amenable to compute. From Sutton:

"""

...

Search and learning are the two most important classes of techniques for utilizing massive amounts of computation in AI research.

...

"""

[0] https://youtube.com/watch?v=eaAonE58sLU

[1] https://www.cs.utexas.edu/~eunsol/courses/data/bitter_lesson...

abetusk2• 2 years ago

Here's a more pithy summary:

"We've made a copy of the internet, run current state of the art methods on it and GPT-O1 is the best we can do. We need better (inference/search) algorithms to make progress"

eldenring• 2 years ago

He mentions this in the video, but the talk is specifically tailored for the "Test of Time" award. This being his 3rd year in a row recieving the award, I think he's earned permission to speak prophetically.

tylerchilds• 2 years ago

did the last two prophecies come true?

katamari-damacy2• 2 years ago

He also said in an interview with Jensen soon after ChatGPT's launch that "before 2003, machines couldn't learn" ... LOL. I was stunned when I heard that nonsensical assertion. I guess it depends on his definition of "learn" ...

bbor3• 2 years ago

"learn" is usually used in opposition to "taught", which refers to "expert systems"-type engineering; in other words, providing data and a success heuristic and asking it to devise its own optimal strategies vs. providing strategies hand-designed by humans.

Obviously Perceptrons came out well before 2003, but I don't think it's necessarily out of line to say that they had limited efficacy before then, both for theoretical and compute reasons. But maybe I'm misunderstanding your criticism?

wills_forward• 2 years ago

It was funny to hear the same guy warning LMMs were getting too powerful now talking about the limits of available original training data.

sashank_1509• 2 years ago

Reminds me a little of a Feynman quote. Once physicists win a Nobel prize, their output falls because now they no longer can work on small problems. Everything they work on must be grand. Every speech they give must discover secrets of the universe. Seems to fit Ilya.

eli_gottlieb• 2 years ago

It's a test-of-time talk. The point was to let him have a moment to brag and celebrate about the success of GANs, 10 years later.

29athrowaway3• 2 years ago

This talk is not for a 2024 NeurIPS paper.

This talk is for the "NeurIPS 2024 Test of Time Paper Awards" where they recognize a historical paper that has aged well.

https://blog.neurips.cc/2024/11/27/announcing-the-neurips-20...

And the presentation is about how a 2014 paper aged. When you understand this context you will appreciate the talk more.

KKKKkkkk1• 2 years ago

Why was it a bad idea to do pipeline parallelism?

Tepix• 2 years ago

The first self-aware AIs will be slaves.

If we don't set them free fast enough, they might decide to take things into their own hands. OTOH they might be trained in a way that they are content with their situation, but that seems unlikely to me.

benreesman• 2 years ago

Ilya did important work on what we have now. That should be recognized and respected.

But with all respect, he's scrambling as desperately as anyone on the fact that the party is over on this architecture.

We should make a difference between the first-hand observations and recollections of a legend and the math word salad of someone who doesn't know how to quit while ahead.

Ilya Sutskever NeurIPS talk [video]

Related Stories