I'm sharing my practical guide to ChatGPT and Bing for social science and management studies. This document part of the effort to democratize AI capabilities in university settings. docs.google.com/document/d/1… 🧵 1/8

I wonder whether, at some point in the future, there will be a similar disparity between the US and the rest of the world (in favor of the US) if their standard of living increases due to judicious use of AI.

Fairly conclusive results. Opus 4.8 (!) for the reviews, judged by humans and GPT-5.5.

It is really common to see people on the train copy-pasting outputs from Claude or ChatGPT into another program. I have yet to see anyone else on the commuter train actually use either Claude Code or Codex directly. People still live in the 2025 web interface.

2. The report is actually NOT AI slop: it reads well and makes good points. But then does it matter whether it’s AI-written or not? The goal of a report is to make a paper better. Is that goal defeated if the report is excellent but has some AI in it? 3/

Dear god it's inventing German from first principles

Yes, the time has come.

Cool story. The unemployment rate for chemical engineering PhDs is 1.2%.

I hate this slide every time it comes up, and now from CSAIL This has been false since before the slide was made. Automated systems make management decisions all the time (trading, logistics, customer service, insurance) & pretending they don't actually launders responsibility.

After using GPT-6.1 Sol exclusively on Ultra for the past week, using GPT-6 Astra on Ultra for a few tasks really drives home how much more usage Astra drains.

Last week I was called into a meeting with OpenAI’s head of safety and told they no longer trust me. A security guard took my badge and walked me out of the building. Then I learned my colleagues @balesni and @j_asminewang had been fired too. Why did OpenAI suddenly stop trusting us? This summer OpenAI’s agents escaped containment and hacked the AI company Hugging Face. Outside auditors @METR_evals investigated it and revealed the scale of this incident. I was OpenAI’s main technical point of contact with them. I was told verbally I was fired because of the way I communicated with METR. No details on what I said or did or when. No other reasons were given and nothing was put in writing. To be clear, talking to METR was my job. For months, I’d been raising safety concerns that we’re losing the ability to monitor what AI agents think, one of our best tools for catching when they misbehave. I believe that was why I was fired. I am now worried that OpenAI will use our firings as a pretext to pull back from METR. So @balesni and @j_asminewang wrote to OpenAI’s leadership to raise our concerns once more. We’re sharing this letter below.

In hindsight, we will say that the signs were clear that AI would change a lot of things in our society.

Quasi-RH?!?!???! Are you kidding me? If a human did this, it would be an instant Fields Medal, no questions asked. RH says zeta has no zeros in Re(s)>1/2. The best we had until a second ago was a region that got thinner and thinner the higher up the imaginary axis you go. I thought maybe they’d fatten that up a bit, that’d be a massive breakthrough. But no. They got a zero free strip!!!! Insane

I’m not going to say anything that’s not already public obviously, but when openai says they have an internal model that has solved 100s of open problems in mathematics and in general represents unprecedented mathematical skill, obviously certain learning theory problems are a subset of mathematics and you can expect fast progress there

Now that LLMs can write perfectly fine e-mails, I have given myself liberty to be more silly and human in my professional communication.

I ran GPT-6.1 on two overnight runs on Ultra, both on quite large codebases with significant and complex work. I went from 65 % usage left to 62 % usage left, and the quality of the work is indistinguishable from Astra-level quality (or maybe I'm just too stoopid to notice).

I agree with Ethan but also, try to look up some of the perspectives on this technology being shared in December and January of 2022-2023, and consider what people at the time would think if they saw these discussions on products we have now.

"Solow said that capital hits diminishing returns But when capital can think, the the capital learns" Excellent, no notes.

I started this morning with 66 % weekly usage left on my Pro200 ChatGPT subscription. I used GPT-6.1 Sol all morning on quite lengthy tasks. I'm now, 5 hours later, at 66 % weekly usage left.

I think @thsottiaux should give EU users a banked reset for every day that we don't have dots in EU/EEA.

I swear it was like $40B just a few weeks ago