I’m also happy to announce another new recruit: James Payor. James joins the MIRI research team after three years at Draftable, a software startup. He previously studied math and CS at MIT, and he holds a silver medal from the International Olympiad in Informatics, one of the most prestigious CS competitions in the world.
- A new paper from Stuart Armstrong and Sören Mindermann: “Occam’s Razor is Insufficient to Infer the Preferences of Irrational Agents.”
- New AI Alignment Forum posts: Kelly Bettors; Bounded Oracle Induction
- OpenAI’s Jack Clark and Axios discuss research-sharing in AI, following up on our 2018 Update post.
- A throwback post from Eliezer Yudkowsky: Should Ethicists Be Inside or Outside a Profession?
News and links
- New from the DeepMind safety team: Jan Leike’s Scalable Agent Alignment via Reward Modeling (arXiv) and Viktoriya Krakovna’s Discussion on the Machine Learning Approach to AI Safety.
- Two recently released core Alignment Forum sequences: Rohin Shah’s Value Learning and Paul Christiano’s Iterated Amplification.
- On the 80,000 Hours Podcast, Catherine Olsson and Daniel Ziegler discuss paths for ML engineers to get involved in AI safety.
- Nick Bostrom has a new paper out: “The Vulnerable World Hypothesis.”