Reproduce “Agent Incentives: A Causal Perspective” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Tom Everitt, Ryan Carey, Eric Langlois, Pedro A Ortega, Shane Legg
The authors use causal influence diagrams to give sound-and-complete criteria for response incentives, instrumental control incentives, and value of control, and prove completeness for a value-of-information criterion. Their examples distinguish potential control from control through an agent’s decision, and show how objectives alter incentives. The criteria describe incentives admitted by an accurate single-decision graph, not guaranteed behavior or incentive strength.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.