Sabitlenmiş Tweet
is/ought
147 posts
is/ought
@telocene
aspiring hunter-gatherer
north sentinel island Katılım Mayıs 2023
187 Takip Edilen19 Takipçiler
@micsolana @david_shane You might say it's more in our interest to lie, but its not feasible. Many people who, like us, believe in ASI, but think its bad (some being employees at frontier labs) will know we're lying and call us out. And, in 2 years capabilities will be so good it will be clear we lied.
English
@micsolana @david_shane @AviWoolf But we will change everything about human existence. We're building a new species far smarter than humans. We think this is very good but will cause short-term strife. Hence, we are honest about what we believe is coming.
English
As @AviWoolf might say, some of this mania might have been avoided if AI companies had stuck with "we are a new and helpful tool" instead of "we will totally change everything about human existence".
Solana is also correct here.
Mike Solana@micsolana
abundance dem defending a moratorium on data centers to appease the communists in his party. as I predicted in my atlantic piece, the far left and the center left cannot coexist, and the center left is cucked. which means the far, regressive left will win.
English
@deanwball The top-left are the people with high P-doom. The top-right (I'd place myself here) are the true accelerationists, and various people at frontier labs. The bottom right is A16Z and co. And the bottom left is the majority of left-wing/academia anti-AI people.
English
@deanwball correct. we basically have a 2D-grid: Y-axis is do you believe in superintelligence and X-axis is how much regulation do you want of AI. The bottom half (no superintelligence) is where most people are, including most "accelerationists", even if they don't realize it.
English
@USWREMichael @AnthropicAI If this destroys anthropic, we lose one of the three companies that can win the ASI race for the US. We also lose an accelerant to the rest of the ecosystem. It is simply not pragmatic (regardless of whether its legal). Winning the ASI race is the *only* thing that matters.
English
Oh @amir from Dis-Information, the Supply Chain Risk designation under 41 USC 4713 is in full force and effect. The USC 3252 preliminary injunction is on hold for 7 days while on appeal. At this moment, @AnthropicAI is a Supply Chain Risk under both sections of the statute.
Amir Efrati@amir
Not a surprise, this judge just gave Anthropic a preliminary injunction, which stops Pentagon from following through on supply chain risk designation
English
@paulnovosad No. all responses to Dario, are simply bets against ASI. if you believe AI gets a bit better but we don't get true ASI, yes, current jobs might disappear but new jobs, best done by humans, will emerge. If you believe we get ASI, then AI does everything better than humans
English
All of these predictions assume nobody else in the system is ever going to change their behavior.
He is right that a tiny share of entry-level people will be doing exactly the job they would have been doing in 2021.
This is not called unemployment.
CG@cgtwts
Anthropic CEO: “50% of all entry-level Lawyers, Consultants, and Finance Professionals will be completely wiped out within the next 1–5 years." grad students and junior hires are cooked.
English
@sp6runderrated i assure you dario actually believes this. as does sam. go read sam's blog posts from 2014 on super machine intelligence. or read any of the lesswrong posters from early 2010s. was all of it just some gigabrain play to raise funds 15 years later.
English
I'm very skeptical this is true despite believe AI will raise structural unemployment.
Tech has a hype culture. I'm basically fine with it. But if your hype culture involves making rhetoric like this you will lose public esteem.
CG@cgtwts
Anthropic CEO: “50% of all entry-level Lawyers, Consultants, and Finance Professionals will be completely wiped out within the next 1–5 years." grad students and junior hires are cooked.
English
@javakai28 @tenobrus @TheLastNeocon @GalvinAlmanza no. the point is that anything that humans can do AI will be able to do far better. so there will be new things to do but AI will do said new things better. the only jobs where humans will still have outright advantage are those where being a human is core to the job.
English
@tenobrus @TheLastNeocon @GalvinAlmanza Guy who thinks that AI capabilities will expand exponentially but the total economy will remain exactly the same even as productivity increases tenfold
English
I don't get how people are planning to sidestep the very basic problem that if you don't have junior hires right now, you won't have experienced people 5 or 10 years later.
CG@cgtwts
Anthropic CEO: “50% of all entry-level Lawyers, Consultants, and Finance Professionals will be completely wiped out within the next 1–5 years." grad students and junior hires are cooked.
English
@tenobrus @HowlingToad @GalvinAlmanza so, they become superhuman pattern matchers even while training on human data (or really any data which isn't pure noise and requires info extraction). and, well, a superhuman pattern matcher is just superhumanly intelligent.
English
@tenobrus @HowlingToad @GalvinAlmanza my view is that they are basically learning multiple distributions but in order to do that efficiently they actually learn *how to learn distributions*, and to do this efficiently they just spend lots of compute learning how to pattern match...
English
@litcapital @AOC yes ofc. but what's the marginal increase to truth/hedging/liquidity provided by 22 y/o college men trading these markets relative to only institutional investors being involved, and does the (IMO, small) marginal benefit outweigh the cost. i'd argue no.
English
@AOC Prediction markets are very helpful in identifying the truth and measuring probabilities of outcomes (like your near 0% chance of becoming president in 2028)
English
This is sad. I know as a politician these companies are going to spend a billion dollars against me for saying it but 🤷🏽♀️
Pervasive gambling is not good for society. It turns life into a casino, traps people in addiction & debt, surges domestic violence, and fosters manipulation.
Polymarket@Polymarket
We’re honored to announce MLB has named Polymarket as their Exclusive Prediction Market Exchange Partner. Polymarket 🤝 MLB
English
@BayesianNuance @Empty_America 11/11
there still is a weak drop off to the right but the point at which said drop off occurs is right shifted—e.g. maybe 6'7 instead of 6'3.
so yea... that's basically my theory
English
@BayesianNuance @Empty_America 10/n
played out over time, this process drives most women's applied value function to look very different from their original value function for height. basically the curve is shifted massively to the right with an additional harsher drop off in value to the left
English
Until very recently, there was a sort of stigma attached to excessive height.
"Tall guys" were often dorky, uncoordinated, bad posture, skinny, etc.
Even in the 90s, most guys over 6'1" were not "cool" at all.
nativistconcern@nativistconcern
In the 1950s, 5ft8 men had more initial marriage success and reproductive success than 6ft2 men. That's not to say they were more attractive, but society functioned. Female mate selection is now largely about impressing other women, looking good in internet photos and chasing media created fantasies. These evo-psych professionals have every justification for women making our species extinct but never mention the real social and psychological dynamics.
English