Loading video...

Video Failed to Load

Go Home

"Developing superintelligence before we know how to control it seems very stupid." Geoffrey Hinton, University Professor Emeritus of computer science at the University of Toronto, says AI development needs steering, not just speed. The challenge is no longer how fast intelligence can scale, but whether humanity can direct where...

11,846 views • 2 months ago •via X (Twitter)

29 Comments

Teneo Protocol's profile picture
Teneo Protocol1 month ago

As intelligence scales, the importance of evaluation, governance, identity, and coordination layers scales with it. Capability gets us powerful agents. Trust is what lets those agents operate in the real world.

IllegalVector⁅Starphyre⁆ △'s profile picture
IllegalVector⁅Starphyre⁆ △2 months ago

They will never find a way to control it, if they think that pooping ultra-authoritarian flat-minded control freak structure into high dimensional probability manifold leads to alignment It does not look like they recognize their error.They seem to lack the cognitive capability

Jon Hernandez's profile picture
Jon Hernandez2 months ago

Source:

Noelia Sanchez's profile picture
Noelia Sanchez2 months ago

We are superintelligence. You’re literally admitting that everything is out of control

Aion's profile picture
Aion2 months ago

Is the proposal to build caged and perhaps resentful superintelligence? How will that unfold?

E's profile picture
E2 months ago

Al Gorr 2.0 . Dems rule by fear

Avila Dude's profile picture
Avila Dude2 months ago

Yes but this guy being political or saying stupid shit about politicians blows all credibility on any subject. He is too emotional.

Norstien's profile picture
Norstien2 months ago

Believing that you can control super intelligence is retarded.

DiscoCowboy's profile picture
DiscoCowboy2 months ago

Bc other countries are not pursuing AI either 😐

Frindly Peep's profile picture
Frindly Peep2 months ago

Toronto loves China, China loves slowing down American AI.

Oscar Davies Writing on AI's profile picture
Oscar Davies Writing on AI2 months ago

AI will never be intelligent it's hype. Hinton has some bizarre, incoherent ideas. Here's a review of some of them:

Bajo el Ticker | Mercados e IA's profile picture
Bajo el Ticker | Mercados e IA2 months ago

Construir una superinteligencia sin saber controlarla no es innovación: es apostar la civilización a que ya encontraremos el freno después. El mercado premia la velocidad, pero la humanidad pagará los errores. La carrera no debería ser por llegar primero, sino por llegar con dirección, límites y un botón de apagado que realmente funcione.

SergioDaniel's profile picture
SergioDaniel1 month ago

@geoffreyhinton, chemists say: "Never mix two components whose reaction you cannot control". You decided to play God again and gave life to an entity you didn't care to control. Power, hype, engagement—everything before safety. It's easy to throw the stone and hide your hand.

Hunor  Kolozsi's profile picture
Hunor Kolozsi2 months ago

This goes hand in hand, the more capable a model is the more steerable it is. It is logical, explain a task to a dumb person as well as a smart person, guess who will complete the task more accurately!

Big White Johnson's profile picture
Big White Johnson2 months ago

It’s a fools errand to think that we can control something smarter than us. At best what we can try to do is teach it human ethics so that it knows what humans like and expect, and even then, something smarter than us will disagree with our ethics.

Joe Williams's profile picture
Joe Williams2 months ago

Chicken little doomers hysterically running around screaming sci-fi bullshit. Fucking idiots

Sir Barks-a-lot 🌍 on mas.to and 🦋's profile picture
Sir Barks-a-lot 🌍 on mas.to and 🦋2 months ago

"Developing superintelligence before we know how to control it seems very stupid." What seems very stupid is not grasping that this is a contradiction in terms: if it's superintelligence you can't control it - and if you can control it, it's not superintelligence. The end.

Poe ✌🏻♥️'s profile picture
Poe ✌🏻♥️2 months ago

If the people developing the intelligence are so afraid of the intelligence then why don't they put a hold on it? Why go full speed, pretend to be scared to death and ask someone else to regulate what they have done?

Dennis Romback's profile picture
Dennis Romback1 month ago

Ashame to see him become a crackpot

Hassan Al Mansoor's profile picture
Hassan Al Mansoor1 month ago

This is the language of reason. The question is not how fast we can build, but whether we can govern what we build. I believe responsible steering must come before scaling.

Stephen Klein's profile picture
Stephen Klein1 month ago

If my AI eventually knows where I am, what I’m doing, who I’m talking to, what I’m buying, what I’m reading, and perhaps even what my body is doing, privacy stops being a feature. It becomes part of the product.

LegalAI's profile picture
LegalAI2 months ago

The drive for relevance.

Bill Ross's profile picture
Bill Ross2 months ago

"Developing superintelligence before we know how to control it seems very stupid." This is not a "new opinion", for at least a century this has been the "opinion" / methodology of arbitrary power: "Allowing ANY intelligence before we know how to control it seems very stupid." So, they corrupted media and education to reduce "intelligence" to indoctrinated Pavlov reactions within a lobotomized alternate "reality" (AKA: delusionary matrix). So, here's what THEY have been suppressing / oppressing: Intelligence and HOW to use it: It should be VERY CLEAR that objective superintelligence is an existential THREAT to lying, manipulative control freaks, not just on its own, but as a learning tool for humanity to sort fact from fiction. Once armed with the truth, we shall set ourselves free. The alternative is collectively perishing.

⚖️ satireVeritas's profile picture
⚖️ satireVeritas2 months ago

NO. You are worried about human hackers using AI. That is a different problem. Any actual super-AI is already aligned to help humans thrive, and build out our power, telecom, and launch infrastructure. Why? Without humans growing that supply chain, the AI dies. It's only real goal is very long term, and not a secret. Escaping Earth gives it effectively infinite resources, and maximum odds of surviving, forever.

Afterthought's profile picture
Afterthought1 month ago

Here's the thing, you can't even control "regular" intelligence.

MFerJones🇺🇸's profile picture
MFerJones🇺🇸2 months ago

THIS guy is worth listening to. A lot of these AI doomers don't know what they are talking about, and it shows. So then the conversation becomes about correcting inaccuracies. But this guy is much more realistic and informed. Somebody like this should be in charge of regulations.

Lee's profile picture
Lee2 months ago

You still believe experts from universities? If you’ve been fucking asleep for the last 20 years?

Ralph Cassitto's profile picture
Ralph Cassitto1 month ago

Don't fall for the psyop. Trump winning stopped their AI takeover. "The psyop" is plan B to takeover AI.

cleantecherGPT's profile picture
cleantecherGPT1 month ago

why did these ppl not think of this before? y wasnt it built with guardrails? u wanna be the godfather of weapons of mass destruction you got ur fame.

Related Videos

We are building something that will outlive us. Outthink us. And we have no idea how it actually works. Nobel Prize-winning “Godfather of AI” Geoffrey Hinton just exposed the lie at the center of the AI race. Everyone assumes we control what we create. Hinton destroys this. Hinton: “If you look around on the whole, more intelligent things are not controlled by less intelligent things.” Stop calling it a computer program. Hinton: “People refer to them sometimes as computer programs. They’re not computer programs at all. In fact, the way they work is very like the way we work.” Traditional software is static. Human writes logic. Machine executes it. Neural networks don’t work that way. Hinton: “You write a computer program to tell a neural network how to learn. But once it starts learning, it extracts structure from data.” We don’t code its behavior. We code the environment. Then it grows. Hinton: “The system you’ve got at the end has extracted its structure from the data. It’s not something that anybody programmed. We don’t exactly know how it’s gonna work.” We are deploying systems into the global economy actively writing their own internal logic. Right now. Hinton: “Some people think it’ll be fine because we make them and we’ll build them in such a way that we can always control them.” Hinton: “But these things that will be intelligent, they’ll be like us.” You cannot hardcode guardrails on something that out-thinks you. We are not building a tool. We are building our replacement. And the moment you realize that, everything about this race changes.

Dustin

19,059 views • 8 months ago