Reading circles

One paper at a time. We pick something that has actually shaped the field — Ngo et al. on the alignment problem, Carlsmith on power-seeking AI, Amodei et al.'s concrete problems — and give everyone a week to read it. The session is two hours: roughly thirty minutes of summary for people who didn't finish, and ninety minutes of argument. We try to separate 'I don't understand this' from 'I disagree with this' — both are valid but they lead to different conversations. Open to everyone; no technical background required for most sessions. We meet biweekly.

• One paper, circulated one week before
• Open to all, no background required
• 2 hours: summary + discussion
• Biweekly cadence
• Annotated reading notes shared after