Trevor Levin
3K posts
Trevor Levin
@trevposts
Trying to help the world navigate the potential craziness of the 21st century, currently via AI Governance and Policy at @coeff_giving
DC usually, SF sometimes 가입일 Haziran 2013
1.9K 팔로잉3.3K 팔로워
Claude is not able to write validly about decision theory... and I would be shocked if EA grantwriters could spot its actual mistakes. The day you're fired is not when you can't see the difference in AI outputs, but when your boss's boss can't see the difference.
🐝StabbithaAllAlong🐝@Stabbitha2
I keep saying that AI *sounds good* but if you ask it to demonstrate in an area of personal expertise, you can see how much bullshit it's really offering. Like, IDK how good the coding is, but I can extrapolate from it's art skills. 😬
English
@RichardMCNgo @slatestarcodex That's probably right, but I think the moloch essay is like <<<1% of the intellectual base of the OAI/Ant race
English
@trevposts @slatestarcodex A world where Microsoft races Amazon is very different from a world where OpenAI races Anthropic. Structurally speaking neither of the former two are *capable* of believing in AGI, even if their leaders sometimes say they do.
(DeepMind racing Meta is somewhere in the middle.)
English
@RichardMCNgo @slatestarcodex Nvm, rest of the thread wasn't loading for some reason. I guess I'm much more of an economic determinist than the thread and think the AGI race was basically baked in by Moore's Law (though not as fully determinist as some)
English
@RichardMCNgo @slatestarcodex I'm very confused by this tweet. Do you really think a world without ACX is meaningfully less likely to have AGI companies than our current world? Or are you not making a casual claim here?
English
The world's most consequential Molochian dynamic is the race between AGI companies.
But several of the leading companies were significantly influenced by @slatestarcodex (e.g. as below).
So perhaps we should think of Scott as having *summoned* Moloch.
Sam Altman@sama
the blog post I think about most often: slatestarcodex.com/2014/07/30/med… (other than a bunch of PG essays, which I somehow don't count as blog posts)
English
Trevor Levin 리트윗함
This tweet got over 1M views so we made it a video:
How much money does Meta make by enabling crimes?
"Internal docs leaked to Reuters show:
• 10% of all Meta revenue comes from ads for scams & banned goods ($16B/year)
• Meta estimates it's involved in 1/3 of all successful scams in the US
• That suggests they drive $50B in scam losses for US consumers alone each year
• Meta earns ~$3B annually from scam/banned goods ads run by Chinese operations alone..."
Rob Wiblin@robertwiblin
Latest podcast from @Gregory_C_Allen has an insane section on criminal activity at Meta. Internal docs leaked to Reuters show: • 10% of all Meta revenue comes from ads for scams & banned goods ($16B/year) • Meta estimates it's involved in 1/3 of all successful scams in the US • That suggests they drive $50B in scam losses for US consumers alone each year • Meta earns ~$3B annually from scam/banned goods ads run by Chinese operations alone The China case study: • In 2024, Meta made $18B+ from Chinese companies advertising to foreign consumers • Internal teams found ~19% was scams/banned content • An anti-fraud team successfully cut these ads in half • When Zuckerberg saw the revenue impact, he told them to "pause" and the team was disbanded • By mid-2025, banned ads climbed back to 16% of China revenue • This results in money being stolen and going directly from ordinary Americans to Chinese criminals The deliberate enabling: • Fraud earns 10% of all revenue, but anti-fraud teams were blocked from any action costing >0.15%, so they couldn't effectively do anything • Meta charged higher rates for suspected fraudulent ads — a "scam tax" • Their algorithm naturally identifies people vulnerable to frauds and feeds them more and more The cold calculation: • Meta anticipated up to $1B in regulatory fines for this • But they make $3.5B every 6 months from high-risk ads • They view these fines as just "cost of doing business" Senators Blumenthal & Hawley now calling for FTC/SEC investigations in a blistering letter, noting that all this happened while Meta cut safety staff and moved billions over to VR and AI. WTF.
English
@trevposts @NateSilver538 Only a tiny fraction of those games starts out with a forecast as extreme as 99%-1%, though!
English
If a #1 seed falls 10 points behind any school that runs a polling institute at any point in the game (Siena, Quinnipiac, Marist) they should just be eliminated IMO.
English
@NateSilver538 Doesn't this show the model is underconfident? >60 games a year, should have plenty of 1% upsets if calibrated?
English
Would have been the biggest upset ever btw. We actually have Norfolk St. vs. Missouri (~2%) in 2012 ahead of the two 16-1 upsets as the least likely historical NCAA outcome, but Duke-Siena would have been a 1% chance.
English
Trevor Levin 리트윗함
This is a pig who spent years confined in a gestation crate. She was left behind when factory farms flooded in Iowa and then rescued by some volunteers.
The volunteers took her home and dug her a mud pit. She ignored it. They assumed years of confinement had extinguished her natural instincts.
Then they noticed her wandering into the woods on their property. They followed her — and found her rooting in a pile of dirt, digging her own mud pit.
The pork industry claims pigs adapt to confinement. They don't. Inside every gestation crate is an animal who still yearns to root, wallow, and just be a pig.
This is the tragedy of factory farming. We tried to reduce feeling animals to machines. We failed.
English
@shhhbreslin @mjnblack They're talking about how they came off as weird and/or endearing to the reporters and laughing about specific lines that had that effect, seems completely normal?
English
there's a truly bonkers hot mic moment at the end of this that may change the way you think about anthropic
you're gonna want to read all the way through this one
vanityfair.com/news/story/dar…
English
@mjnblack I don't think they said anything that should make one think this.
English
@trevposts I have people tell me all the time that their doomer comms strategy is just marketing (i.e. the david sacks perspective)
I've always given them the benefit of the doubt, but this was the first time I wondered if I was in fact being played
English
Trevor Levin 리트윗함
In a large upset, LTF (the OpenAI-Andreessen super PAC) takes a major loss in IL-02, where they backed Jesse Jackson Jr. Notably Jackson is famously corrupt, and I wonder if LTF’s toxic AI money fed into existing negative sentiments towards him.
English
Trevor Levin 리트윗함
Meta's $45 million super PAC lost 3/4 of the Illinois state legislature races it spent in last night.
Meta spent $750,000 across four state races in Illinois and their only chosen candidate who won was Paul Kendrick in IL-HD-12.
English
@epiplexity Totally agree, this is the world we live in, and incremental progress matters a lot. Just sad about it.
English
@trevposts Like what am I going to do, complain that the sky is blue?
No, you just work to make 90% of the supply chain suffering free and help bootstrap lab grown meat.
English
It makes me really deeply sad about human nature that this post has 100k likes.
Angelic.Pomegranate@RealFoodRina
What are vegans going to do when they find out plants have consciousness
English
Trevor Levin 리트윗함
• According to the story, the dog's cancer has not been cured.
• Absent all regulatory and manufacturing constraints, we could not just synthesize magic mRNA cancer cures. The technology is very promising, but it's not yet any kind of panacea.
• The emergent system of regulators and manufacturers is indeed far too conservative, and small-scale experimentation is much harder than it should be. More people should read the first part of The Rise and Fall of Modern Medicine. Recommend @RuxandraTeslo, @PatrickHeizer for more.
English
@BogdanIonutCir2 Are mats extensions not ~this? Scaling stuff like that seems like it could absorb tens of millions but not the quantities people are usually discussing
English
@trevposts e.g. possible bypass: fund lots of independent researchers (e.g. grad students) for 1 year, evaluate after 1 year whether to renew; this doesn't seem to require managers/leads, though it might require more grant evaluators - and not even necessarily, if less risk-adverse
English
I keep seeing this discourse about how there aren't enough great things to fund in AI safety and Idk where it's coming from. When I was doing field building, there didn't seem to be any scarcity of talented people trying to join in; instead, scarcity/fear of funding, like today.
English
@AlexanderMcCoy4 Campaign finance world has this very dumb fixation on the "win-loss record" of super PACs, making the obvious strategy to back likely winners and oppose likely losers. That's probably why LTF hasn't opposed Scott Wiener despite him advancing very similar legislation
English
While the spending from AI PACs is an important part of the story in NY-12, I think it’s important for journalists and pundits covering the outcome to not oversimplify, whatever the result ends up being.
I think that Lasher was always the most likely to win based on the fundamentals of the district (in West Side vs East Side, the West Side generally wins because there are more votes there, & Lasher has the blessing of the outgoing Rep), and now Lasher has Bloomberg spending for him.
Also important to note that while Bores has been an AI Safety champion, and thus is somebody big tech donors in SF have heard of, Lasher was also a co-sponsor of all the same AI Safety legislation Alex Bores was.
Quite frankly, this is a bit of a strange race to turn into a referendum on AI policy, for all those reasons.
Jeff Coltin@JCColtin
Pro-Micah Lasher mailer arrives from Stand up for New York PAC, just days after the Times reported that Mike Bloomberg will be funding it with millions of dollars. New stage in the #NY12 race - most previous super PAC spending was attacks on Alex Bores.
English