Introducing Directionally bad
A weekly newsletter on AI centralization (and the people fighting it)
We’re launching Directionally bad: a weekly newsletter on the centralized AI labs quietly redrawing the map of who gets access to intelligence — and a stubborn few trying to build an open alternative.
Here’s why we’re doing it, and what you’ll get if you subscribe.
Get Sultan Suleimanov’s stories in your inbox
Join Medium for free to get updates from this writer.
Directionally bad is published by DeAI Nation, a global non-profit working to explore and analyze Decentralized AI. As AI takes over the world, accessing it at a reasonable price is getting harder and harder. Huge AI labs are securing millions of GPUs — the essential hardware behind AI training and serving — and dictating access to their models, which are built on data scraped from the internet, books, and us, without consent.
A small group of outcasts is trying to build an open, decentralized alternative. Open-source code, open-weight models, and distributed compute are the crucial blocks of Decentralized AI. If they succeed, people around the world will have a better option than locking their data, processes, and finances into one of the existing centralized labs. If they fail… well, you’ll be the first to know — from our newsletter.
Directionally bad goes out once a week, on Saturdays. We cover major moves from centralized labs, progress in the Decentralized AI community, and the data points you need to monitor the state of compute access. All for free.
If any of that sounds like a thing you’d want to read on a Saturday morning, drop your email below. And if you know someone — a friend, a colleague, an AI agent — who should be watching this space too, send it their way.
See you Saturday.
