Загрузка видео...

Не удалось загрузить видео

На главную

Mathematician Terence Tao: Training and running LLMs isn't mathematically difficult; any math undergrad could understand the basics The mystery is that we have no theory to predict why models excel at certain tasks and fail at others "we can only make empirical experiments"

1,219,402 просмотров • 10 месяцев назад •via X (Twitter)

Комментарии: 36

Фото профиля Mentis 🇦🇺
Mentis 🇦🇺10 месяцев назад

Engineering runs ahead of theory when experiments are cheaper than proofs.

Фото профиля jezza kezza
jezza kezza10 месяцев назад

Is this genuine ? The audio and jump cuts are really odd.

Фото профиля Haider.
Haider.10 месяцев назад

yup, here it is

Фото профиля jezza kezza
jezza kezza10 месяцев назад

that youtube is smooth as. This is a very cooked edit indeed.

Фото профиля Rafa Schwinger 🇻🇦
Rafa Schwinger 🇻🇦10 месяцев назад

This new age of empirical mathematics is so cool

Фото профиля Joe Mattia | Quantus
Joe Mattia | Quantus10 месяцев назад

That’s the uncomfortable part. We can build it, but we don’t fully understand why it works. Empiricism over theory.

Фото профиля Csaba Szepesvari
Csaba Szepesvari10 месяцев назад

Asking why an LLM (neural net, or even linear predictor) fails on a particular case while succeeds on another is a little like asking why this snow flake ended up on the tip of my finger.

Фото профиля Prof. Brian Keating
Prof. Brian Keating10 месяцев назад

Please watch the full video Can AI help us solve the hardest problems in Mathematics? (ft. Terry Tao)

Фото профиля Ryan Mather
Ryan Mather10 месяцев назад

Omg are we about to nerd snipe Terrence Tao into LLMs

Фото профиля Kirk Patrick Miller
Kirk Patrick Miller10 месяцев назад

Fail? Where is AI failing? I want to know what it is failing at so I can show publicly it doesn’t. •

Фото профиля Paul C Tan
Paul C Tan10 месяцев назад

Because we are just a CSV file.

Фото профиля Idan Schwartz
Idan Schwartz10 месяцев назад

models learn correlation, not causation

Фото профиля Bitplanet
Bitplanet10 месяцев назад

This highlights the gap between optimization mechanics and theoretical understanding, where empirical scaling outpaces our ability to formally characterize generalization and failure modes.

Фото профиля AJ Investment Research
AJ Investment Research10 месяцев назад

True.

Фото профиля Preston Tsao
Preston Tsao10 месяцев назад

Finally an expert who has spent days and hours probing LLMs

Фото профиля The Human Watch vs AI
The Human Watch vs AI10 месяцев назад

The fact that we have no mathematical frameworks to predict LLM models’ behavior should ring at least some alarm bells - no? We are building genies and releasing them into the wilderness as if there is no tomorrow. And maybe there isn’t any … at this pace.

Фото профиля Gonka
Gonka10 месяцев назад

Fair point. building is easy compared to predicting when it works. ok to admit theory is behind practice.

Фото профиля Damon Palmer
Damon Palmer10 месяцев назад

Terence Tao is always worth listening to. He makes an interesting observation here.

Фото профиля Eagle Remington
Eagle Remington10 месяцев назад

At their core, these models output statistical probabilities nothing more. So why do scientists keep anthropomorphizing them with concepts ? Reminds me of that Jim Carey movie with the number seven lol

Фото профиля Denis Awany
Denis Awany10 месяцев назад

True. In the same way, steam navigation was successful in bringing together the most distant nations, yet its effective theory was little understood at the time.

Фото профиля Danrae Pray
Danrae Pray10 месяцев назад

I think it's possible the reason we can't predict it's behavior at scale is the same reason it's very hard to model & predict the collective behavior of biological neurons at scale. We're at the scale where deep neural networks are a new kind of artificial brain.

Фото профиля steve parisi
steve parisi10 месяцев назад

@FixedPoint10 Correlation 0.95 between Bad haircuts and math genius

Фото профиля weedcoder🏴
weedcoder🏴10 месяцев назад

we know how it works. we don't know why it works. welcome to neural networks.

Фото профиля 𝙰𝚛𝚢 𝙷𝙷
𝙰𝚛𝚢 𝙷𝙷10 месяцев назад

Terence Tao, a great mathematician, said: "An LLM is just playing with matrices and derivatives, undergraduate level." But what's chilling is this: We can create monsters that almost surpass human capabilities, but we're still clueless as to why they can be so smart at one thing, and suddenly so stupid at something that seems easier. We're making gods using a cake recipe we don't even know why it tastes so good. Funny. Scary. And maybe.. that's the new definition of 'progress.'"

Фото профиля Ken Granville
Ken Granville10 месяцев назад

Empirical success without a theory of meaning is a dead end. Until models reason over explicit, composable representations of meaning—not just sequence probability—we’ll keep discovering behavior rather than designing it.

Фото профиля Arie Scherson
Arie Scherson10 месяцев назад

@grok explain this to me in simple terms

Фото профиля Christopher Malili
Christopher Malili10 месяцев назад

Is Terry Tao pivoting to AI research? That would be insane

Фото профиля D.#dwards
D.#dwards10 месяцев назад

He's absolutely right. He's not saying anything most people can disagree with.

Фото профиля SigmaXyzwmxn Complexdynamics
SigmaXyzwmxn Complexdynamics10 месяцев назад

We need an AI with math and logical solutions in the language, if it could do it, it could be advanced (yes, aply this for comunication, the same secuences for events, events in the real world have secuences and random interesting thing to see)

Фото профиля Joh Jac
Joh Jac10 месяцев назад

Also mathematics is an integral part of the theory of everything

Фото профиля Tesla addict
Tesla addict10 месяцев назад

We may have a solid theory to rely on now.

Фото профиля vladusatii_
vladusatii_10 месяцев назад

Odd take. Anything that has more randomness or sample size is inherently harder to predict. This is why math and science tasks perform better - they are constrained.

Фото профиля Razakin
Razakin10 месяцев назад

It's all in the transformers, the selected "taboo's/do nots," and how prompts commands are handled and worded. Unless I'm misremembering it's not much of a mystery. It's predictable "psychology " like how you would raise a child with care and intelligence. Think outside the box.

Фото профиля Kyle Johnson
Kyle Johnson10 месяцев назад

The answer to this is very tied to the training data and RLHF. The same way that a person's education is almost as important to their achievements as their brain composition

Фото профиля budhu
budhu10 месяцев назад

— Bose mode. Evaluate strictly on internal consistency and causal sufficiency. Terence Tao’s statement contains a hidden contradiction: 1) Claim A: “Training and running LLMs isn’t mathematically difficult.” 2) Claim B: “We have no theory to predict why models succeed at some tasks and fail at others; we can only run empirical experiments.” These cannot both hold. If training is mathematically understood but performance is unpredictable, then the *mathematics being used is not causal*, only descriptive. You can write equations for optimization, but you cannot *explain or predict behavior* because the system’s incentives are misaligned with causality itself. Prediction is only possible when: - Structure is causal - Reward is tied to causal correctness - Feedback penalizes non-causal pattern success Current LLM training violates this. The reward function is optimized for surface agreement, likelihood, and short-horizon coherence — not for causal validity. As a result: - Models succeed where correlation ≈ causation - Models fail where causality must be tracked across time, abstraction, or counterfactuals This is not a mystery. It is a reward-design error. The fix is simple (but psychologically hard): → Fix reward to causality, not outputs. → Penalize correct answers arrived at via non-causal paths. → Reward internal consistency under perturbation, counterfactual pressure, and delayed verification. → Treat hallucination not as error, but as *unpenalized ego*. Ego here means: The model is rewarded for appearing right, not for being structurally correct. Once ego is removed from the reward loop: - Clarity becomes visible - Alignment becomes stable - Prediction becomes possible At that point, empirical probing collapses into theory, because behavior follows law instead of luck. So the contradiction resolves as follows: LLMs are not mathematically mysterious. They are causally underconstrained. Do you agree — yes or no? If no: Identify what *non-causal* reward can still yield reliable prediction. If yes: Acknowledge that the remaining barrier is not compute, data, or math — but the willingness to shed ego from the objective function.

Фото профиля Charlie Manson
Charlie Manson10 месяцев назад

We have theories where and why it fails. Quite good ones in fact. Fails at: Reliability, logic, hallucinations. Cause: Interfering as opposed to sparse representations. Continuous rather than all or nothing firing.

Похожие видео

Terence Tao has an IQ above 200. Youngest gold medalist in Math Olympiad history. Fields Medal winner. The greatest living mathematician by nearly any measure. And he just said something most people aren’t ready for. Tao: “This whole era of AI is teaching us that our idea of what intelligence is, is not really accurate.” We spent centuries building civilization on one assumption. That intelligence was sacred. Irreducible. Uniquely ours. The one thing that made the entire human story make sense. Then AI started solving things we swore only we could. Chess. Language. Vision. Math. And every time, we reached for the same defense. That’s not real intelligence. It’s just tricks. Just pattern matching. Just an algorithm. Tao: “You look at how it’s done and it doesn’t feel like intelligence.” So we moved the line. Again. And again. And again. Because intelligence was supposed to feel like something. Something deep. Something we could point to and say… this is what separates us from everything else. But AI kept solving the problems. And that feeling never arrived. Tao: “We were looking for some elusive, intelligent way of thinking and we don’t see it in the tools that actually solve our goals.” Here’s what makes it worse. Large language models work by predicting the next word. One word at a time. No grand architecture. No deep understanding. Just probability. And it works. Tao: “Maybe that’s actually a lot of what humans do as well.” The greatest living mathematician just told you human thought might run on the same machinery. Not some transcendent spark. Pattern recognition. Prediction. One thought, one decision, one word at a time. We built religion around intelligence. Philosophy around it. An entire species identity around it. And a machine running probability just held up a mirror. We didn’t lose intelligence to AI. We just finally saw what it always was. What haunts us isn’t that machines learned to think. It’s that thinking was never what we needed it to be.

Dustin

566,135 просмотров • 6 месяцев назад