Meta's Vision for Superintelligence • Hacker News Board

Re: Meta's Vision for Superintelligence

#101

Post by simonsarris » Wed, Jul 30, 2025, 3:20 PM UTC

sailingparrot wrote: post #58

> We'll need to be rigorous about mitigating these risks and careful about what we choose to open source. Here we go, predictably pulling the oldest trick in the book, just two weeks after it was reported [1] that the Superintelligence leadership was discussing moving to closed source for their best models, not for any risk mitigation reason, but for competitive reasons. Also, > As recently as 200 years ago, 90% of p…

> Here we go, predictably pulling the oldest trick in the book, just two weeks after it was reported [1] that the Superintelligence leadership was discussing moving to closed source for their best models, not for any risk mitigation reason, but for competitive reasons.

That's not pulling a trick, that's doing precisely what Zuck said he would do. In April 2024 Zuck on Dwarkesh said that models are a commodity right now, but if models became the biggest differentiator, that Meta would stop open sourcing them.

At the time he also said that the Model itself was probably not the most valuable part of an ultimate future product, but he was open to changing his mind on that too.

You can whine about that anyway, but he's not tricking anyone. He has always been frank about this!

Re: Meta's Vision for Superintelligence

#102

Post by criddell » Wed, Jul 30, 2025, 3:20 PM UTC

alazoral wrote: post #98

criddell wrote: post #53

Earlier quoted context omitted.

I define it as smarter than humans. The same way a dog will never understand L'Hôpital's rule, there are probably things that human beings will never understand but future AIs will. I'd call those AIs superintelligent .

I think it’s too easy to get anthropocentric and 1D about intelligence. That dog understands things about the world, important things, that mere humans like Bernoulli could never even dream of.

It's anthropocentric because it's being built by humans. If a dog built a superintelligent AI, I would assume it would be canocentric.

Re: Meta's Vision for Superintelligence

#103

Post by AIPedant » Wed, Jul 30, 2025, 3:23 PM UTC

criddell wrote: post #53

khelavastr wrote: post #4

How is "superintelligence" defined?

I define it as smarter than humans. The same way a dog will never understand L'Hôpital's rule, there are probably things that human beings will never understand but future AIs will. I'd call those AIs superintelligent .

It is worth emphasizing that dogs wouldn't understand formal calculus even if they were smarter than humans! They are not physiologically capable of doing formal mathematics:

- their eyesight is too poor to read

- their paws are not designed for fine manipulations so they cannot write or type

- their throats and mouths are not nearly as nimble as ours, so they cannot vocally communicate detailed information

Even if there was an Newton-level dog, they wouldn't be able to access the ideas of an earlier Euclid-level dog. Human knowledge is not just about our big brains, we've developed many physical features that make transmission of information far easier than other species.

OTOH dogs do have a good intuitive "common-sense" understanding of arithmetic, geometry, and physics. It is the unique gift of humans that we can formalize and then extend this intuition, but this ability (and intelligence as a whole) relies on nonverbal common sense.

Re: Meta's Vision for Superintelligence

#104

Post by 9rx » Wed, Jul 30, 2025, 3:28 PM UTC

lvl155 wrote: post #97

9rx wrote: post #94

Earlier quoted context omitted.

> I am not sure if investors will continue to give him a blank check. What are they going to do, exactly? They explicitly invested in the company knowing that Zuckerberg would retain full control. If they can show gross negligence there may be a legal avenue, but it would be pretty hard to justify chasing potentially profitable business ventures, even if they end up failing, as being negligence. Controversial busines…

Well, very true. They let him spend $100B plus on a mediocre VR devices that’s not improved all that much under Zuck. I keep thinking, what if they took that $100B and invested in Nvidia.

Hindsight is 20/20, I suppose. Not much they can do now, though, except sell their interest to someone else. But why would someone else want to buy their interest when that someone else could use the same money to buy a stake in Nvidia instead?

Re: Meta's Vision for Superintelligence

#105

Post by fredoliveira » Wed, Jul 30, 2025, 3:35 PM UTC

blibble wrote: post #27

SpicyLemonZest wrote: post #12

Earlier quoted context omitted.

Anthropic also expects that superintelligence will be reached in the somewhat near future. The intuition is that there's just not that much distance between a chatbot that can manage a vending machine poorly and a chatbot that can manage it well.

that intuition is wrong

In what ways?

Re: Meta's Vision for Superintelligence

#106

Post by YetAnotherNick » Wed, Jul 30, 2025, 3:35 PM UTC

AIPedant wrote: post #103

criddell wrote: post #53

Earlier quoted context omitted.

I define it as smarter than humans. The same way a dog will never understand L'Hôpital's rule, there are probably things that human beings will never understand but future AIs will. I'd call those AIs superintelligent .

It is worth emphasizing that dogs wouldn't understand formal calculus even if they were smarter than humans! They are not physiologically capable of doing formal mathematics: - their eyesight is too poor to read - their paws are not designed for fine manipulations so they cannot write or type - their throats and mouths are not nearly as nimble as ours, so they cannot vocally communicate detailed information Even if t…

All of these is true for Stephen Hawking. Of course one can argue that he couldn't have done it if not for other humans making technology for him.

Re: Meta's Vision for Superintelligence

#107

Post by Gud » Wed, Jul 30, 2025, 3:43 PM UTC

benterix wrote: post #89

FirmwareBurner wrote: post #65

Earlier quoted context omitted.

>There's a bunch of people who want a large income (wealth) disparity Apart form you of course, so I'm sure you'd be ok if the government would tax your higher than average tech wage till your take home pay would match that of a train conductor's or bus driver's, like in Western Europe, and therefore fix the wage gap you hate so much. Would you like that solution? Caption this: It's only a problem when the people who…

> the government would tax your higher than average tech wage till your take home pay would match that of a train conductor's or bus driver's, like in Western Europe I live in Europe and earn ca. 6 times more than my friend who is a bus driver in the same city. We both have access to free education and, if we wish, also free healthcare, for which I am paying slightly more, but I really don't mind.

If you earn 6 times more than your friend who is a bus driver, you live in a place that has an unusually high income disparity for Europe.

Re: Meta's Vision for Superintelligence

#108

Post by criddell » Wed, Jul 30, 2025, 3:45 PM UTC

AIPedant wrote: post #103

criddell wrote: post #53

Earlier quoted context omitted.

I define it as smarter than humans. The same way a dog will never understand L'Hôpital's rule, there are probably things that human beings will never understand but future AIs will. I'd call those AIs superintelligent .

It is worth emphasizing that dogs wouldn't understand formal calculus even if they were smarter than humans! They are not physiologically capable of doing formal mathematics: - their eyesight is too poor to read - their paws are not designed for fine manipulations so they cannot write or type - their throats and mouths are not nearly as nimble as ours, so they cannot vocally communicate detailed information Even if t…

> It is worth emphasizing that dogs wouldn't understand formal calculus even if they were smarter than humans!

Exactly! If a dog invented a dog superintelligence and it discovered calculus, the dogs would never understand that discovery. I think a superintelligence we build will discover things we cannot understand.

Re: Meta's Vision for Superintelligence

#109

Post by AIPedant » Wed, Jul 30, 2025, 3:50 PM UTC

YetAnotherNick wrote: post #106

AIPedant wrote: post #103

Earlier quoted context omitted.

It is worth emphasizing that dogs wouldn't understand formal calculus even if they were smarter than humans! They are not physiologically capable of doing formal mathematics: - their eyesight is too poor to read - their paws are not designed for fine manipulations so they cannot write or type - their throats and mouths are not nearly as nimble as ours, so they cannot vocally communicate detailed information Even if t…

All of these is true for Stephen Hawking. Of course one can argue that he couldn't have done it if not for other humans making technology for him.

The big difference with Stephen Hawking is that he was not born disabled, he became disabled during graduate school. Even in 2025, a human who is born blind and severely paralyzed (so they cannot speak or sign) will probably never learn calculus, regardless of innate ability. Perhaps in the medium term technology will improve.

That said, another major difference is psychology. Switching animals, it seems plausible to me that chimpanzees are theoretically capable of doing basic calculus as a matter of pattern-matching. But you can't force them to study it! Basic calculus is too tedious and high-effort to learn for a mere banana, you need something truly valuable like "guaranteed admission to the flagship state university" for human children to do it. But we don't have an equivalent offer for chimps. (Likewise an Isaac Newton - level dog might still find calculus exceptionally boring compared to chasing squirrels.)

Re: Meta's Vision for Superintelligence

#110

Post by AIPedant » Wed, Jul 30, 2025, 3:57 PM UTC

criddell wrote: post #108

AIPedant wrote: post #103

Earlier quoted context omitted.

It is worth emphasizing that dogs wouldn't understand formal calculus even if they were smarter than humans! They are not physiologically capable of doing formal mathematics: - their eyesight is too poor to read - their paws are not designed for fine manipulations so they cannot write or type - their throats and mouths are not nearly as nimble as ours, so they cannot vocally communicate detailed information Even if t…

> It is worth emphasizing that dogs wouldn't understand formal calculus even if they were smarter than humans! Exactly! If a dog invented a dog superintelligence and it discovered calculus, the dogs would never understand that discovery. I think a superintelligence we build will discover things we cannot understand.

No, you totally misunderstood my point. Formal calculus is not a physical fact of the universe, it is an abstract human tool that other species (including dogs) are not capable of using, just like they can't use hammers or drive cars. The problem is not a lack of intelligence, it's having the wrong body.