musings
In October 2026, an AI agent solved 49 of 50 unpublished problems from 11 specialties of theoretical physics, after its objections sent the experts back to find errors in their own answers. Philosophers call it epistemic trespassing when an expert passes judgment outside their field. As AI makes it cheap to work in a field you never trained in, the ability to judge that work becomes the scarce part of science.
About a thousand high-school students in Turkey practised maths with GPT-4. Their practice scores rose 48%, and then they scored 17% lower on the exam than classmates who had used only a textbook. Routine work used to make two things at once, the output someone paid for and the expert nobody paid for, and AI can now deliver the first without the second.
In 1847 a Vienna doctor named Ignaz Semmelweis made medical students wash their hands, and deaths among new mothers in his clinic fell by almost 90%. Two years later he was out of a job. Stanford professor Jeffrey Pfeffer argues that doing good work and getting people to adopt it are two different skills, and that the second one can be learned.
Reid Hoffman says social networks do best when they tap into one of the seven deadly sins. The sins explain why someone signs up. They explain much less about what happens to that person afterward.
Xiaoyin Qu predicts that AI will drive the value of every task with a checkable answer to zero, leaving profit only in matters of taste. When answers get cheap, the premium moves to choosing the questions and checking the answers.
An AI agent might find the same rental for less on a host’s site. The hard part is proving that the offer is real and the booking is safe. That is where Airbnb’s fee faces its test.
Patrick Collison expects personal AI agents to act as a structural subsidy for product quality. Agents will raise the reward for quality only where quality can be verified independently of the seller.
Some people get lucky more often than others. Most of the difference comes down to noticing more and knowing people outside your usual circle.
Of the NeurIPS papers one committee accepted, a second committee rejected 50.6%. One rejection is a noisy observation.
Persistence pays only when it is aimed at the right thing. Commit to the problem, and treat each solution as a hypothesis you are ready to drop.
Most people choose a role and inherit its problems. Choose the problem first, then find the role that lets you work on it.
Lottery winners were barely happier than people living nearby. Ambition runs into the same problem. Each achievement quietly becomes the new baseline.
A high salary and a high-leverage career are not the same thing. The difference is how closely your earnings are tied to your own time.
My MSc advisor Joëlle Pineau took the title of her commencement address from bandit algorithms. An action is worth taking not because its estimated reward is high, but because the uncertainty around it leaves room for it to be much better.
Anthropic is the notable exception among the major labs, and critics see protectionism from a company whose business depends on keeping its own models closed. Both interpretations can be true.