We're Learning Backwards
LLMs are brilliant at many things, but still fail in unreasonable ways. Is that an issue of scale, or something more fundamental?
Please Don't Cite Me
Musings of the nerd-sniped. Things worth understanding deeply, and things that probably aren't.
Latest posts
I Waited 10 Billion Cycles and All I Got Was This Loading Screen
Modern hardware is absurdly fast, which mostly seems to have bought software the freedom to feel slower, heavier, and less deliberate than it has any right to.
HTTP/1.1 500 Segfault: Building a web server in C
Native OAuth eventually turns into a question about local process boundaries, and answering it cleanly means learning just enough HTTP to build a tiny localhost server in C.
The Ethics of Computer-Generated Content
From image generators to code autocomplete, generative AI is reshaping what it means to create. But as adoption accelerates, our existing legal and ethical frameworks are not sufficient.
Reinforcement Learning Adventures
Teaching neural networks to play Chrome Dino, Connect 4, and Othello.