Conversation menu #

This is a non-exhaustive list of topics I’d be keen to discuss. Better yet, let me know what’s been on your mind lately. Email is good.

AI risk #

• I assume AI risks fall into either one of three categories: deliberate misuse (authoritarian regimes using AI for value lock-in, careless deployment of AI systems by private individuals), misalignment (the technical AI safety problem) and social disruption (risks associated with mass unemployment, other uncertainties). My worry distribution looks something like this: deliberate misuse 0.5, misalignment 0.1 and social disruption 0.4. What’s yours?
• A working hypothesis: “There are two AI safety research directions with disproportionately high impact: first, safety and capabilities evals of existing models; second, adapting cybersecurity defenses to an internet populated by AI agents.” Is this right?

AI capabilities #

• It seems to me like the main benefits of AI might come from accelerated scientific progress. Can AI beenfit ordinary people more directly? I think the answer is yes.
• Just how neglected is AI literacy? I’m starting to view this as the biggest obstacle between humanity in Feburary 2026 and a hypothetical AGI utopia.

Serious matters #

• What do you think of bullshit?
• How does one become more agentic?
• Which books have influenced your worldview the most?