How very salesman of Altman.
The Intelligence Age
361–370 of 447 posts
Re: The Intelligence Age
#362Earlier quoted context omitted.
"me when i need $6.5B" https://x.com/netcapgirl/status/1838265672096243822
Where is $6.5B from?
Whomever buys into it
https://www.bloomberg.com/news/articles/2024-09-11/openai-fu...
Re: The Intelligence Age
#363> humanity discovered an algorithm that could really, truly learn any distribution of data (or really, the underlying “rules” that produce any distribution of data) He's hand-waving around the idea presented in the Universal Approximation Theorem, but he's mangled it to the point of falsehood by conflating representation and learning. Just because we can parameterize an arbitrarily flexible class of distributions doe…
> It's about as profound as the notion that you can approximate a polynomial by splicing together an infinite number of piecewise linear functions. Wait 'til you hit complex analysis and discover that Universal Entire Functions don't just exist, they're basically polynomials.
So basically Altman hasn’t taken enough math courses yet thinks he is starring in Good Will Hunting
Re: The Intelligence Age
#364Just as the previous one was the information age, right? Which was a lie.
Re: The Intelligence Age
#365> It is possible that we will have superintelligence in a few thousand days (!) "a few thousand days" is such a funny and fascinating way to say "about a decade" Superficially, reframing it as "days" not "years" is a classic marketing psychology trick, i.e. 99 cents versus a dollar, but I think the more interesting thing is just the way it defamiliarizes the span. A decade means something, but "a few thousand days" f…
"me when i need $6.5B" https://x.com/netcapgirl/status/1838265672096243822
this person has amazing shitposts...good account
Re: The Intelligence Age
#366> If we want to put AI into the hands of as many people as possible, we need to drive down the cost of compute and make it abundant (which requires lots of energy and chips). If we don’t build enough infrastructure, AI will be a very limited resource that wars get fought over and that becomes mostly a tool for rich people. Translation for the rest of us: "we need to fully privatize the OA subsidiary and turn it into…
Why can't we just go with Occam's razor, and assume that they really believe in their mission of providing access to frontier AI as freely to the world as possible?
Why would that be where Occams Razor leads us?
Re: The Intelligence Age
#367Earlier quoted context omitted.
It's amazing how he can imagine wars being fought over AI but not wars being fought over resources needed to "build enough infrastructure."
> he can imagine wars being fought over AI I wonder how did he come to such a prediction. When was the last time we had a war over advanced tech? Armies didn't fight over telegraph, radio, phones and cars. A war to get AI would also be foolish. A few hundred bombs from your adversaries and AI won't have electricity to function.
Telegraph, radio, phones, cars didn't guarantee success. Those things can't outthink you. Somewhat extreme case: let's say Israel saw that Iran was days away from getting AGI. It would instantly throw every nuke it had at Iran.
Now, scale that up to the global superpowers and near-superpowers. I wouldn't want to put money on what e.g. the U.S. would do if China was days away from AGI. There are a lot of options below going nuclear, e.g. assassinations, hacking, destruction of equipment.
Once someone has a thing that can invent, hack, assemble, strategise 24/7, replicate itself, all bets are very much off. We need to sleep, it doesn't.
Re: The Intelligence Age
#368> humanity discovered an algorithm that could really, truly learn any distribution of data (or really, the underlying “rules” that produce any distribution of data) He's hand-waving around the idea presented in the Universal Approximation Theorem, but he's mangled it to the point of falsehood by conflating representation and learning. Just because we can parameterize an arbitrarily flexible class of distributions doe…
"Just because we can parameterize an arbitrarily flexible class of distributions doesn't mean we have an algorithm to learn the optimal set of parameters."
This is equally mangled, if not more, than what Altman is saying. We don't need to learn "the optimal" set of parameters. We need to learn "a good" set of parameters that approximates the original distribution "well enough." Gradient methods and large networks with lots of parameters seem to be capable of doing that without overfitting to the data set. That's a much stronger statement than the universal approximation theorem.
Re: The Intelligence Age
#369Earlier quoted context omitted.
Labor organizers mostly make things worse for laborers. Capitalism has brought much prosperity. "It is not from the benevolence of the butcher, the brewer, or the baker, that we expect our dinner, but from their regard to their own interest"
> Labor organizers mostly make things worse for laborers Nonsense, unless you're prepared to argue that the existence of weekends is "much worse" than working every day.
Labor unions in America were largely controlled by organized crime and likely still are today. They didn't do this to help the worker they did it as an extension of a collection racket. AKA - their own interests. Despite this they did accomplish some historical good. Yet, I am still working a 40+ hour week bullied by my employer and jobs continue to be shipped overseas. How many more decades before the unions do their job?
You can't continue to give labor unions credit for what they've done in the past as justification for their continued existence. Labor unions are also the reason the police are untouchable even when they commit horrendous crimes. Anecdata, but of the union members I know (who by the way were forced into it lest they be considered a scab) none of them believe the union works for them. These people are in medicine, education, construction, and food. The union rep likely golfs and drinks with the CEO where they agree after "much fighting for your rights" to grant you that 25 cent raise.
Rarely unions do their jobs. During COVID hospitals in my hometown were allowed to bully nurses into dangerous situations as their union would not halt care in order to make sure they had functional N95s. The unions didn't stop the mass exportation of labor to third world nations after the GFC. These things alone are enough for me to say they have outlived their usefulness.
Re: The Intelligence Age
#370Earlier quoted context omitted.
It's amazing how he can imagine wars being fought over AI but not wars being fought over resources needed to "build enough infrastructure."
>resources needed to "build enough infrastructure." Strarship will put into orbit 100tons at $10M or so. I.e. And wars willn't be fought over AI. Wars will be fought using AI. Humans will have no chance in controlling millions of their own and responding to the actions of the millions of the enemy's simultaneously active automated high-precision munitions of all kinds, and that picture leads to a new, non-nuclear thi…
> Wars will be fought using AI.
I agree with the sentiment but not the mechanism (munitions). If power equates to control of resources then destroying the resources is kind of self-defeating. Far better then to "win" a war by getting friendly regimes installed that are aligned to your interests. We're already seeing the importance of this kind of approach, but the ultimate evolution will be that what used to be called propaganda (disinformation) will not just facilitate a war of guns and bombs, it will _be_ the war - able to directly achieve the end-goals of the "war" with not a single shot fired. AI is a huge amplifier of capability in this domain.