In late October 2026 I shared two LinkedIn posts summarizing key take‑aways from my interview with Emmanuel Ameisen, a senior interpretability researcher at Anthropic. The interview was released as a podcast episode sponsored by WorkOS and DX.
Interpretability as a field is small, young, and open to newcomers.
models are "grown," not built, and his job is "much closer to biology than to engineering."
We discuss:
🔸 How interpretability connects to AI alignment and safety