A few discussion points:
1. AI is accelerating. I'll focus on the AI → code → strategy pipeline.
2. Major chunks of "big-picture judgment" can likely be automated soon - possibly including aligned AI leading altruistic strategic decisions.
3. Background from the forecasting world: wins, losses, and why even well-funded efforts (Coefficient Giving is winding down its Forecasting Fund) struggle to find traction.
5. A "robust epistemics" agenda: picking problems that are an unusually good fit for AI oversight.
6. Live demos of two QURI tools — RoastMyPost (multi-agent document critique) and LongTermWiki (LLM-maintained, claim-sourced knowledge base) — with lessons learned.
Few slides; mostly demos and Q&A.