I still haven't heard a good answer to this question, on or off the podcast. AI researchers often tell me, "Don't worry bout it, scale solves this." But what is the rebuttal to someone who argues that this indicates a fundamental limitation?

Dwarkesh Patel tl;dr: Maybe learning simple things (basic knowledge, heuristics, etc) actually lowers the loss more than learning sophisticated things (algorithms associated with higher cognition that we really care about), and the sophisticated things will eventually be learned as scaling

I always wanted to learn to code some basic stuff, but I was extremely lazy about learning syntax, etc. Luckily, thanks to AI, even a smooth brain like mine can turn ideas into indicators without begging 20 other people to help him. I will publish these on Tradingview (for

Book Summary: “Structures”

🧵35 years of fighting the market beast. Here is my path and some things I have realized, but not all of course. I started in Wall Street as a fixed income quant. I wrote algos to optimize bond portfolios based on constraints, and objective functions. 1/n

Okay. Thanks for the nerd snipe guys. I spent the day learning exactly how DeepSeek trained at 1/30 the price, instead of working on my pitch deck. The tl;dr to everything, according to their papers:

In less 60 minutes I built an AI agent that: 1. Crawls a website 2. Looks for organic traffic opportunities 3. Prioritizes actions based on impact This is a microdose dose of what's possible right now. Most marketers have no idea what's about to happen. Insane leverage,

the DeepSeek R-1 research paper is wild. i’m still wrapping my head around the implications of page 8. my highlights below along with a summary of the full paper. (we are in for a wild ride)

LLMs can't solve compositional tasks. In other words, if you learn Category Theory, AI won't replace you.

There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It's possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper

This guy with a ChatGPT subscription is doing substantially better data work than like, 90 percent of data scientists.

It’s already happening

What Andrej Karpathy calls 'vibe coding' isn't just a style -- it's the next fundamental abstraction layer in computing. In 1973, Unix devs sparked outrage by rewriting assembly in C. We're seeing the same reaction to AI-driven development. Here's why this matters 🧵

This guy perfectly explains how humans will think and learn in 2030

If you want to learn how to write mathematical proofs, try this wonderful book for self study. Book of Proof by Richard Hammack. You can order a physical copy or get it as a Free PDF download. Last year I studied the PDF version of this book and it was pure joy.

this is what's keeping me up at night these days... 1. google has become unusable. once you get used to "deep research" (thanks grok, perplexity etc), google feels like bringing a typewriter to a macbook meeting. 2. MCP will do for ai agents what REST did for web services -

This is what y'all should be focusing on!!

today we are introducing codex. it is a software engineering agent that runs in the cloud and does tasks for you, like writing a new feature of fixing a bug. you can run many tasks in parallel.

New blog post where I explain why I disagree with this, and why I have slightly longer timelines to AGI than many of my guests. I think continual learning is a huge bottleneck to the usefulness of these models, and extended computer use may take years to sort out. L-nk below.