NavinF retweeted
Customer B has been revealed. Anthropic is SpaceX's largest single customer, overtaking NASA and the Pentagon for the first time. 34.5% of SpaceX's revenue last quarter was from compute-as-a-service to Dario Amodei, and Anthropic has committed to spending up to $120.5 billion through 2029. He will be paying Elon $3.1 billion a month starting in December 2026.
NavinF retweeted
There’s basically 50 million people paying for the other 300 million people to live, and at every level they’re mocked and derided for it.
NavinF retweeted
ed zitron has made like endless predictions about ai that have been proven black and white WRONG and yet he keeps getting touted as an AI expert
NavinF retweeted
I thought I knew every horrible thing about this story, but it was not until I read @neerajadeshp's interview with Sabatini that I realized how total his unpersoning was
Former MIT scientist David Sabatini just won $6 million in a landmark case against false allegations of r*pe and sexual assault, which destroyed his career.
Once on track to win the Nobel Prize, Sabatini’s ex-girlfriend accused him of r*pe in 2020 and began calling him the ‘Weinstein of science.’ He was later accused of perpetuating a “bro culture” in his lab. As a result, he was stripped of all his funding, lost his job, and went bankrupt.
We interviewed Sabatini (@DMSabatini) about the verdict, which he says made him “proud to be an American.”
@neerajadeshp reports 👇
NavinF retweeted
New definition of “Chad” just dropped
Trump appointee btw
New: A lawyer’s misconduct complaint against US District Judge Ada Brown stemmed from a hearing during which she repeatedly directed racial insults at him, according to documents reviewed by Bloomberg Law.
news.bloomberglaw.com/us-law…
NavinF retweeted
One of the reasons the New York Times stays on top is that you’re reading this thoroughly reported article about Dem Party malfeasance not in the pages of the Free Beacon or the Daily Caller but from the Grey Lady herself.
SCOOP: Democrats have insisted they have nothing to do with the other Dan Sullivan on Alaska's Senate ballot.
But aides with the Mary Peltola campaign and DSCC were part of a monthslong secret plan to recruit him to run -- as well as other candidates.
nytimes.com/2026/10/28/us/po…
NavinF retweeted
I think there's something of an implicit appeal being made by Weisenthal and other "Save the Bars" supporters which I agree with and am willing to state explicitly:
It's nice to have places which were built before the tyranny of neurotic safetyism, the ADA, and plastic.
NavinF retweeted
I feel like one thing that's not really captured in all the benchmarks is how ridiculously fast GPT models in codex are at doing what you want compared with Claude. It feels like magic when I switch from claude agents to Codex.
NavinF retweeted
> BUT THIS LIKELY MEANS THE LIMITS ARE MERGED AS WELL
yep seems that way. Haven't been using the app/CLI at all, only web chats (with Astra Pro), yet going down by 1% every 5 minutes.
Web cheese is over, despite the BS in the app. OpenAI has moved to Anthropic limit model.
ChatGPT Web updated for me, mering chat and work. Now I can choose Astra and 6.1 as chat models, which is cool, and they also have Fast mode. I only hope it won't be 30 tps.
BUT THIS LIKELY MEANS THE LIMITS ARE MERGED AS WELL :( what about Pro mode? Will I lose like 5% of my weekly quota on 1 request?
Big L :(
NavinF retweeted
Okay, this makes so much more sense now.
Open weights models fall further behind in v5. The leader, Kimi K3.1 Pro (weights not yet released), scores 46, 12 points behind the leader, compared with a 3-point gap in v4.3. Of models with released weights, DeepSeek V4.1 Flash 1017 (max) and GLM-5.3 (max) score highest at 32, and Claude Haiku 5.5 (max) outscores all of them.
The gap is widest on HLE-Diamond and Terminal-Bench-Science, where Kimi K3.1 Pro scores 37% and 34%, compared with 68% for Claude Opus 5.5 (max) and 63% for GPT-6 Astra (max).
NavinF retweeted
When AI art took off, it was artists who led the backlash. Now that AI code has taken off, who is leading the backlash? Yep, artists.
NavinF retweeted
The first four boxes don't me much; the final one is eyepopping.
On first four-- wealth and income measurement is *extraordinarily* hard. Good measures of income often come from tax data, but tax data is itself endogenous to the tax regime, etc.
But that last box. 97%. Jeez.
Not sure what is more striking—how bad the research audited here seems to be, or the thought that a lot of other research might come out similarly shaky.
Via @tylercowen / @MargRev – saez-zucman-audit-replicatio…
NavinF retweeted
What was stunning was their willingness to put their disdain for Asian applicants *in writing* and to even try to game the system by having admissions committee members rate Asian personality scores as "low" when the interviewers who had actually met them rated them as "high"
NavinF retweeted
> Gemini 4 Carbon, our next model
because of course carbon is the next model. Now, when they released Argon, I was like "ah, they'll name them after noble gases" and that seemed reasonable. Helium, Neon, Argon, KRYPTON, Xenon, Radon. All beautiful names, and very intuitive, heavier/scarier is more powerful, Helium is a super cheap classifer.
But apparently they didn't want to do that, because google cannot name or brand a product if their lives depended on it. So maybe they're going across the period then, right? Chlorine, Sulfur, Phosphorus... idk, not great, but they'd be in theme at least.
NOPE! The new model is named Carbon. What does Carbon have to do with Argon? Fucking nothing, that's what. Because fuck you. None of it matters. It'll be in your google drive plan and you're gonna use it because you don't care, and so google can name it by just throwing a dart at a table because nobody on their marketing team shows up to work. COOL!
I did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model.
Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching.
We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards.
As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!
NavinF retweeted
just heard someone say "people are more important than the economy" and it occurred to me that some people genuinely believe that others care about the economy because they like numbers going up and not because it drives living standards for the population
NavinF retweeted
seems reasonable to call the decision to not release GPT-2-XL safetyist slop separate from whether it made you an AGI-believer-elect or not in 2019. It was obviously true that the model wasn’t dangerous in any way; even the threats cited at the time were about spam
Dario reveals the people who called his GPT-2 fears "pseudoscience" in 2019 are now all applying for jobs at AI labs
"I personally was super freaked out by [GPT-2]. I looked at it and I said, my God, you just scale this thing up another factor of a million and it'll be better than humans at everything. This is the defining economic and scientific moment of the modern age. We just don't know it yet. This is going to determine the fate of nations and economies and the future of humanity."
"Maybe 10% of the people around me had exactly the same reaction, and it was often the smartest ones. 90% of the people around me, because GPT-2's capabilities in any particular area were not that impressive, it could translate languages, summarize a document and write a few lines of code here and there. It was the broadness and the generality."
"But people would look at it and say, in any particular area, that doesn't seem so smart. And so many people just shrugged and they just dismissed it."
"I just look at it and I say it's the same thing over and over again. The limitations of the models, six months from now, the things that seemed hard, they're going to be easy. 12 months from now, the things that seemed impossible, the models are going to be doing them regularly."
"It's everything we saw with GPT-2; we're seeing it over and over again in wider and wider swaths of the world. And the world is gradually finally being convinced."
Daniela Amodei: "I remember at the time there were some very well-respected computer scientists and AI and ML leaders who basically bullied some of the researchers who worked on [GPT-2] on Twitter and were like, this is stupid. We don't know why you guys are even talking, this model is so incapable."
"It was like you were sort of a quack if you believed this was a really significant moment."
Dario: "That's all been memory-holed. All the researchers who were like, oh, this is pseudoscience, it doesn't make sense, all of them are applying for jobs at the LLM companies."
"We never say it out loud to any particular one, but we look at them and they're like, we remember what you were saying in 2019. You have no credibility on this matter."
