Chris Olah on what the hell is going on inside neural networks
Humanity made these amazing and ever-improving tools. So how do our creations work? In short: we don’t know.
80000hours.orgHere are the latest publicly reported updates about Chris Olah.
- Chris Olah remains a prominent figure in AI interpretability, widely recognized for his work on mechanistic interpretability and his leadership role at Anthropic. TIME named him to the TIME 100 AI list in 2024 for his influence in the field.[7]
- In 2025–2026, Olah has continued to appear in interviews and talks discussing interpretability and safety in large language models, including appearances on long-form podcasts and public lectures that recap his career across Google Brain, OpenAI, and Anthropic.[2][3]
- Public sources describe his ongoing involvement with Anthropic and related interpretability research initiatives, with media coverage noting his role in connecting safety-focused research to scalable model development.[1][4]
If you’d like, I can pull more detailed excerpts from recent interviews or point you to specific talks and publications from 2024–2026. I can also summarize key themes from his recent talks, such as how mechanistic interpretability is being applied to production-scale models.
Citations:
- Olah's TIME 100 AI recognition (2024).[7]
- Public discussions and career notes including Google Brain, OpenAI, and Anthropic (2023–2025).[3][1][2]
- Anthropic-related interpretability work and public communications (2025–2026).[4]