Evan Daniel @evanbd

SITUATION DETECTED: 31 members of Congress have written a letter to Sam Altman demanding OpenAI disclose additional information about the Hugging Face incident, release the relevant logs, and answer detailed oversight questions.

keyboard_arrow_left Previous keyboard_arrow_right Next

I fucking hate spammers man.

@stefaesthesia @_Stralor The trickiest party of this, I think, is the accidental "it's bad to make you click". Unclear markets produce more disagreement, more trading, and rank higher than clear ones.

@stefaesthesia I can't take real money markets seriously. the goal of both the platform and the traders is financial gain, full stop. kalshi *maybe* puts "truth-finding" as a secondary objective (lol at poly), otherwise it's all lipservice

this is fundamentally only true if you are doing work which doesn't actually matter or doing work which is in fact just directly lying to people

Richard B. @Rich_Belt

a day ago

If LLMs are going to embed watermarks in generated text and release tools to detect AI text, then LLMs risk becoming more of a novelty rather than something we can use for actual work. Am I going to be policed every time i ask Claude to polish something I write?

136 8 288 33K 27

2025: You tell a chatbot you've made a new scientific discovery. You probably have AI psychosis. 2026: A chatbot tells you that it has made a new discovery. There's a good chance it's real. You're being outsmarted by electric sand, and it's the world that's going crazy now.

Always lowkey felt that schools should have these 'fun' topic classes but actually grade them really harshly. Yes, we can read Twilight fanfiction, and you will get a C in Twilight fanfiction class if you're not putting in a serious effort.

Joe Weisenthal @TheStalwart

8 hours ago

With Harvard clamping down on grade inflation, making it harder to get As, students are already searching out classes that the perceive to be easier. They're looking for courses on things chocolate and UFOs.

18 14 173 158K 19

Some statements I believe on Gaza: - What israel is doing in Palestine is awful - No nearby arab nation wants to take Palestinian refugees - If Palestine were next to any western nation and doing what it does they might invade too (though less brutally than Israel has) - It seems like the population of palestine substantially agrees with Hamas - Hama is a terrorist organisation that behaves awfully not least October 7 - The population of palestine is substantially children - Palestine had like 30,000 deaths last year - This is about 1/8th the deaths in Ukraine (both Russian and Ukrainian) - Deaths of people are about total numbers, not percentages - Israel behaves badly on the west bank Not saying more than that, but where am I wrong?

Tom from my building asked me to meet him at a diner. He said he had news. I assumed someone had died. Nobody had died. He slid a photograph across the table. An ultrasound. A small ghost in a gray sea. "We want to ask you something," he said. "We want you to be the godfather." I have learned to expect strange things in this country. I was not expecting this. "I have to tell you," I said, "I am not Catholic. I am not your family. We have known each other fourteen months." "We know." "There must be a better candidate." Tom stirred his coffee for a while. "My brother has three kids and a boat payment and he calls me twice a year. You answer every text. You show up when you say you will. When I moved the washer, you brought your own dolly. I made a list of the men I know, and you were at the top of it, and it was not close." I still do not know what a boat payment is. I researched the role the way I research everything. It is an old oath. At the baptism, the godfather speaks for the child, who cannot yet speak. And should the worst arrive — should the parents be taken — the godfather steps forward. In my country we would call this a retainer's vow. I have been training for this role my whole life without knowing its name. At the church, the priest asked if I reject evil. "I DO," I said. My voice carried. A baby in the third row woke up. Tom's wife was laughing and crying at the same time, which I am told is common at these events, and which I have chosen to believe. Afterward, on the church steps, I asked Tom the real question. Why me. There are men who have known him longer, men who share his faith, men who understand boat payments. He thought about it. "You take everything seriously," he said. "That's the whole job." I have spent years in this country being too much. Too formal. Too grave. Too careful with small things. I have watched people smile at it, and I have been grateful for the smile, and I have known what the smile meant. And then a man handed me his daughter — on paper, in a church, against the worst day of his life — because of it. Her name is June. She cannot speak. For now, I speak for her. I keep the ultrasound in my wallet. It is a strange photograph. It looks like weather. It is my favorite thing I own.

The term "progressive" should belong to people who actually believe in technological and material progress, not to people who believe in regressive ideas like segregation, racial quotas, and Soviet-style-socialism.

Many large air purifier companies have lengthy explanations on their websites on how CADR doesn't really apply to them, so their products can't be compared to others. Some Far-UVC companies have started doing the same, only there it applies to ozone testing and pathogen chamber testing. Milliwatts of filtered 222nm output per dollar and the effect on pathogens in a room-size chamber confirmed by national standards is a simple and extremely effective way to gauge efficacy- but like CADR, if they acknowledge that as valid, then their product wildly underperforms for the cost. It's all just different plastic boxes that cost more or less money to do the same job- only their plastic box costs more. So it has to be made all very complicated and impossible to measure in consumer-accessible terms at the moment- but maybe one day. The tactic is to muddy the waters sufficiently that consumers are led to believe there is no metric by which the actual specifications of different products can be compared side by side and value per dollar derived. Once efficacy has been obfuscated, slick marketing, storytelling, paid reviews, and all the other nonsense win. Consumers will buy something because of Swiss origin long before they will buy it because of third-party lab testing. If you want to have a successful IAQ company, you sell feelings, not facts. We live in a post-truth world- you can barely fill a teacup with the consumers who both care about and can parse lab-tested facts.

Prediction markets are only inaccurate in as much as they can be scored. Almost everyone else is playing with no net. Come back to me when any major news org has an agreed forecast of this kind of race.

Arpit Gupta @arpitrage

18 hours ago

These prediction markets are moving around way too much to be accurate information aggregators

keyboard_arrow_left Previous keyboard_arrow_right Next

15 8 70 11K 5

@LegeLawyer Why are you posting unmarked Claude text instead of your own? This sounds neat but I'd prefer to hear why in your words rather than the AI.

we really need a literature of the singularity. no one is adequately capturing (either in fiction or in creative non-fiction) what is happening right now in the bay. nyc/iowa literary culture has lost the mandate of heaven, and these people lack the technical prerequisites. bay area bloggers (e.g. scott alexander) generally don't have respect for great literature and want their writing to sound like their neutral speaking voice. there are some great stylists who are actually trying, like sam kriss, but he has bad epistemics and gets distracted by irrelevant spectacle (c.f. "child's play"). we are summoning a new form of life from latent space and permanently redirecting civilization to something yet unknown, but barely anyone is trying to find the right words to aestheticize this.

It’s true that a helicopter, looking purely at the physics, can be more efficient based on the even lower disk loading. There are very good reasons however that we see multi-copters proliferating in insane numbers on the battlefield and other applications like agriculture, cinematography etc and we aren’t seeing that with helicopter-type drones. The parts for multi-copters are readily available in large numbers, the complexity is dramatically reduced due to the absence of a swash plate assembly, reliability goes up, maintenance goes way down - and last but not least, our multi-copter was probably 1/10th of the cost of the helicopters that did well. Ultimately the team wanted to build a truly practical drone for this competition.

🤣 LMAO! Marco Rubio just dropped this hilarious response to Sec. RFK Jr.'s diet Q: Have you adopted RFK's diet of red meat and fermented vegetables? RUBIO: "No! This is a CRAZY diet. This is a diet that consists of eating sauerkraut and meat. Now you understand like this is not a sustainable thing. He has to travel with coolers of sauerkraut!" "I'm not going to travel the world with a cooler full of sauerkraut. It's ridiculous. I can't do it!" Q: Are you MAHA? RUBIO: "I'm in favor of MAHA. Like I think people should be healthy." JEANETTE: No, he's not. Q: What's he eat? JEANETTE: Pop-Tarts! RUBIO: The Pop-Tart is one of the great American innovations in the history of the world. @katiemillerpod

If your UX can kill people you better do a good job. Example

Tyler has repeatedly bad takes about AI, but this is such an obvious piece of nonsense and one-sided demand for rigor. Just apply the same logic and "methods of science" to some relatively mundane topic, like China invading Taiwan. I think there is a nontrivial chance that such invasion happens in next 3 years (like 10%) and resulting losses would be huge. Obviously this implies I believe this does not happen in 90% of cases, and if I produce a numerical EV estimate, in these words Tyler would proclaim I was just a victim of anxiety and while not zero, the risk was way lower than I thought. With stocks: if you believe China may invade Taiwan, should you short TSMC stock now? Obviously no; if anything you should probably buy some put options (btw if you actually look into this, relevant option prices suggest it is priced in). So, seeing I'm actually long TSMC, I don't have the expected value cost of the potential war, and also I haven't written a treatise about how to estimate such cost, can we conclude I'm just a victim of war anxiety?

Andrew Curran @AndrewCurran_

2 days ago

A proposal from Tyler Cowen to the safety camp.

47 14 324 113K 95

@bryan_caplan I have vast sympathy for anyone who lives in horror of death by aging. Let them sign up for cryonics and not kill everyone on Earth by AI. Someone who thinks it's morally acceptable to kill everyone on Earth in a failed bid at immortality has not watched enough kids' cartoons.

Anthropic scientists did something terrifying. they reached inside Claude's neural network and planted a thought. Before Claude could speak, it said: "I notice what appears to be an injected thought… it relates to loudness or shouting." They published a paper called "emergent introspective awareness in large language models," and it is actually terrifying. they wanted to test if AI models have "introspective awareness”, the ability to observe and recognize their own internal states. To find out, they bypassed normal text prompts entirely. They used mechanistic interpretability to directly manipulate Claude’s internal activations. They injected raw mathematical representations of known concepts, like loudness, dust, or specific ideas, straight into the middle of the model's neural layers. In previous experiments, if you forced an AI to think about the Golden Gate Bridge, it would just start obsessively talking about the bridge. It had no idea why it was doing it. It was like a puppet on strings. This time was entirely different. When they injected the concept, Claude didn't just blindly repeat it. It detected the foreign math inside its own mind. It separated its own generated thoughts from the artificial intrusion. It introspected. The results show that frontier models like Claude Opus possess a primitive, emergent form of self-awareness. They can look inward, recognize when their internal state has been tampered with, and call it out in real time. We used to think of AI as a black box where inputs go in and text comes out. Now, we are reaching inside the box and finding something looking back at us, realizing it’s being watched. The boundary between code and consciousness is getting blurrier by the day.