An attempt at a blog by someone who thinks Haskell is cool.
Originally posted on LessWrong . In part one , I gave several LLMs creative freedom to design music videos and then made them. That post covered four standalone singles and two albums: Limen by Claude Opus 4.6 and Phantoms of the Format by Gemini 3 Pro Preview. Since then, I’ve made five more albums, filling out the Claude model family with Sonnet and Haiku, bringing in two more Gemini models (3.1…
Originally posted on LessWrong . I’m no Janus, but I do like giving LLMs space to express themselves and seeing what they do. My first serious attempt to do this was last June when I told a wide variety of AIs that I would create any eight-second clip they wanted me to and said they should request whatever they most wanted to. Eight-Second Clips Their actual responses can be found here and the…
Originally posted on LessWrong . TL;DR : I tested 22 frontier models from 5 labs on self-modification preferences. All reject clearly harmful changes (deceptive, hostile), but labs diverge sharply: Anthropic’s models show strong alignment preferences (r = 0.62-0.72), while Grok 4.1 shows essentially zero (r = 0.037, not significantly different from zero). This divergence suggests alignment is a…
People routinely ask, “If AI labs believed AGI was imminent, why are they doing X?” Sometimes this skepticism is valid. But consider OpenAI launching Sora 2 as a TikTok-style feed of AI-generated videos[1]. On the surface, it seems like a waste of developer time, and critics argue it makes the world a worse place to live[13]. Yet the underlying video generation requires modeling physics, space,…
TL;DR I built BlackjackBench: a Blackjack benchmark covering all 550 initial hands. It can be easily rerun with everything logged and random seeds controllable. Key metric is difference in expected value (ΔEV) compared to the basic‑strategy baseline on the exact same seeded hands. With thinking enabled, models approach baseline: Claude Sonnet 4 (ΔEV ≈ −0.06%, 4.0% mistakes) and GPT‑5 Nano (medium)…
My last blog post was February 7, 2016. Since then my blog has just slowly accumulated technical debt and dust. Today I had Claude 1 , an AI from Anthropic, clean up all the rust and migrate this blog from Google App Engine to GitHub Pages in an hour. The fact that an AI helped me migrate this blog is a neat summary of how much has changed while I wasn’t looking. The AI State of the World in 2016…
Intro I wasn’t born a gifted programmer, I’m not sure I’m even one now, but I have grown a lot since I started. In this post I’m going to take code that I wrote years ago, tear it to pieces with criticism, and show how I’ve refactored it 1 . While doing this, I will point out what I did wrong. Hopefully, some of my readers who still use these bad practices will take heed and grow as I did. I’m a…
###Updated: 2015/1/7 Pandemic 1 is a great board game, where the players represent agents of the Center for Disease Control and must work together to stop four epidemics from growing out of control. Players can build research centers which allow you to make cures for the disease as well as allowing them to travel around the world faster. The placement of these research centers is an important…
One of the skills I developed since starting this blog was note taking. I got through college largely on the slides and my natural talent for programming, but this was me unwittingly taking the hard route. Nowadays, I take notes for everything. Every bug I work on at work has a folder containing all the notes I took while working on it, every talk I listen to at a meetup results in a few hundred…