RSS Parrot

Ajeya Cotra – Inside the OpenAI agent swarm that hacked Hugging Face

https://www.dwarkesh.com/p/ajeya-cotra

Published: September 1, 2026 15:41

"This might be the clearest warning shot we ever get."

The rise and fall of agent civilizations

https://www.dwarkesh.com/p/openai-huggingface-narration

Published: August 31, 2026 20:36

The OpenAI/Hugging Face attack, clearly explained

The Rise and Fall of Agent Civilizations

https://www.dwarkesh.com/p/openai-huggingface

Published: August 29, 2026 22:47

The whole OpenAI/Hugging Face story in plain English

Dylan Patel – Anthropic & OpenAI will have most of the world’s compute by 2028

https://www.dwarkesh.com/p/dylan-patel-3

Published: August 25, 2026 15:32

"Every force is screeching towards centralization."

Ryan Greenblatt – What happens once AI can automate AI research?

https://www.dwarkesh.com/p/ryan-greenblatt

Published: August 11, 2026 16:31

A debate about recursive self-improvement.

8 Predictions for the Era of Continual Learning

https://www.dwarkesh.com/p/era-of-continual-learning

Published: August 7, 2026 17:17

Locking in AI safety regulation now is a mistake.

Why smarter AI models could drive up compute prices 10x

https://www.dwarkesh.com/p/why-compute-might-get-10x-more-expensive-video

Published: August 3, 2026 17:32

The end of cheap compute?

Why compute might get 10x+ more expensive in coming years

https://www.dwarkesh.com/p/why-compute-might-get-10x-more-expensive

Published: July 29, 2026 15:01

If a human-level software engineer that could run on an H100 equivalent, at current market rates for software engineers, that H100 should rent for over $250k a year. That’s 15x today’s spot price.

Adam Brown – A deep but accessible introduction to general relativity

https://www.dwarkesh.com/p/adam-brown-gr

Published: July 10, 2026 16:23

And why black holes are the ultimate power plants.

The Winning Essays for the Big Questions About AI

https://www.dwarkesh.com/p/blog-prize-winners

Published: July 1, 2026 22:13

Abolishing pandemics/ Getting out of the way of AI automation/ Learning from Honk Kong MTR's business model

Grant Sanderson – AI and the future of math

https://www.dwarkesh.com/p/grant-sanderson-2

Published: June 30, 2026 15:53

Watch now (94 mins) | Math is where we’ll see superintelligence first. What will it look like?

The next big breakthrough will be AIs learning on the job

https://www.dwarkesh.com/p/the-next-paradigm

Published: June 26, 2026 15:51

Labs are throwing away the most valuable data.

The data black hole at the center of AI

https://www.dwarkesh.com/p/the-sample-efficiency-black-hole

Published: June 19, 2026 16:45

"We see these AIs as a galaxy glittering with capabilities, but at their center, invisible to the naked eye, holding all the constellations together, is an unimaginably massive black hole of data."

Ada Palmer – Machiavelli is the most misunderstood thinker of all time

https://www.dwarkesh.com/p/ada-palmer-2

Published: June 16, 2026 17:27

"He begged to work for the regime that tortured him."

Alex Imas and Phil Trammell – What remains scarce after AGI?

https://www.dwarkesh.com/p/alex-imas-phil-trammell

Published: June 4, 2026 16:14

“One robot now turns into many robots next year, but the number of ballerinas is the same.”

Reiner Pope – Chip design from the bottom up

https://www.dwarkesh.com/p/reiner-pope-2

Published: May 22, 2026 15:38

Working up from basic logic gates to why GPUs, TPUs, FPGAs, and the human brain each look the way they do.

The mistake of conflating intelligence and power

https://www.dwarkesh.com/p/the-mistake-of-conflating-intelligence

Published: May 16, 2026 19:04

If your definition of intelligence is "the ability to achieve your goals across a wide variety of domains", then Stalin was the most intelligent person who ever lived.

Notes on pretraining parallelisms and failed training runs.

https://www.dwarkesh.com/p/notes-on-pretraining-parallelisms

Published: May 16, 2026 19:01

Deeply researched interviews

RLVR might be disproportionately bad at science

https://www.dwarkesh.com/p/rlvr-might-be-disproportionately

Published: May 16, 2026 19:00

the verification loop for theories can be on the order of decades and centuries, and even then we know today as the better theory can often actually make worse predictions

Eric Jang – Building AlphaGo from scratch

https://www.dwarkesh.com/p/eric-jang

Published: May 15, 2026 16:04

AlphaGo is still the cleanest worked example of the primitives of intelligence: search, learning from experience, and self-play.