Speaker recommendations for the Guaranteed Safe AI Seminars

[Orpheus Lummis]

Hi all,

We're looking for speaker recommendations for the Guaranteed Safe AI Seminars – a monthly series bringing together researchers to advance AI with quantitative safety guarantees. https://lu.ma/guaranteedsafeaiseminars 

Two quick asks:

- Forward this email to someone you think would be good to have on the series

- Interested in presenting? Apply here: https://airtable.com/appYgVucDqD7WkTyy/pagvLGoo9WCTI9ohk/form

The seminars are 1h, half presentation half discussion, usually occurring on the second Thursday of each month, at 13:00 Eastern time. We publish recordings on Youtube.

So far we had:

• Justice Sefas — Synthesizing Gatekeepers for Safe Reinforcement Learning
• Roman Soletskyi — Verifying Global Properties of Neural Networks
• Steve Omohundro — Provable AI Safety
• Rafael Kaufmann — Gaia: Distributed planetary-scale AI safety
• Evan Miyazono — Proving safety for narrow AI outputs
• Charbel-Raphaël Ségerie & Épiphanie Gédéon — Constructability: Designing plain-coded AI systems
• Yoshua Bengio — Bayesian oracles and safety bounds
• Louis Jaburi — Compact Proofs of Model Performance via Mechanistic Interpretability
• Agustín Martinez Suñé — Using PDDL Planning to Ensure Safety in LLM-based Agents
• Tan Zhi Xuan — Engineering Rational Cooperative AI via Inverse Planning and Probabilistic Programming

Questions or referrals: feel free to email directly.

– Orpheus