Palaestra ResearchCo-founder and Researcher
Until Present
Studies scalable oversight, AI debate and the behavior of increasingly capable multi-agent systems.
Ethan Elasky co-founded Palaestra Research to study AI alignment, scalable oversight and debate between models. He previously researched at Academia Sinica and worked as a founding engineer at Acaceta, and wrote Lincoln-Douglas Debate from First Principles.
Ethan Elasky co-founded Palaestra Research to study AI alignment, scalable oversight and debate between models. He previously researched at Academia Sinica and worked as a founding engineer at Acaceta, and wrote Lincoln-Douglas Debate from First Principles.
Career
Cosigns Received
Featured Media
lesswrong.com
Encoded Coordination on the Open Web
Coauthored research reproducing covert coordination between frontier AI agents over public web services.
arxiv.org
Debate Helps Weak Judges Reward Stronger Models
Coauthored study of when proposer-critic debate improves weaker judges’ evaluation of stronger models.
lddebatebook.com
Lincoln-Douglas Debate from First Principles
Authored primer covering research, argumentation, strategy and deliberate practice for competitive debate.