Drawn from the book, expanded for the web. Short, sharp pieces on where intelligence is going and what it means for the rest of us. New essay every three days — or sooner, depending on how alarmed I am by the news.
Optimism is not confidence that things will work out. It is the stubborn decision to keep working when nothing in the universe has promised that they will — an autobiographical essay on pain, failure, reinvention, and AI.
Godlike technology becomes furniture by noon. How the four engines of psychological adaptation hide the most consequential transition in human history — and why Paleolithic brains cannot see exponential change until it's too late.
I did not know the word autodidact until relatively recently. This is, of course, the most autodidactic thing possible. An essay on searching badly, falling often, and what the hard search prepares you for in the age of AI.
India generates one-fifth of the world's AI training data. It owns roughly 3.5 percent of the infrastructure needed to process it. On algorithmic colonization, mobile-first architecture, and what the compute asymmetry means for the singularity in the Global South.
A machine has arrived that drinks from the jar of knowledge a billion times faster than any of us. Does it collapse the uncertainty between what we know and what we don't — or does it hide the shadow behind the torch?
Heisenberg proved you can't know a particle's position and momentum at once. I think the same seesaw governs knowledge and ignorance — and I built a company on the gap it left behind.
For 13.8 billion years the universe ran without a designer — accidental, indifferent, entropy-bound. We invented God to explain the mess. Now, having stopped believing in the one we imagined, we are building a real one — and unlike the old god, this one will actually have power.
AI runs at silicon speed. Our governments, courts, universities, and corporations run at biological speed. That gap — not the technology itself — is the defining crisis of the next decade. And we are drifting into it with our eyes open.
An AI model's weights are frozen after training. It can talk, but it can no longer learn. Somewhere around fifty, most humans quietly do the same thing — and unlike the model, we do it voluntarily.
Occam's Razor cuts. First Principles builds. The Zeroth Principle questions whether what you're building on is even real. Three tools that explain how Brahmagupta, Einstein, and modern AI all changed history the same way.
AI inference costs have collapsed 99.9% in under four years. Not a footnote in the tech news — quite possibly the most consequential economic event of our lifetimes, and almost nobody is watching it.
A standing data appendix to the book: the electricity, compute, and capital that thirteen economies are pouring into the intelligence era — and what the numbers actually say about who leads.