Reproduce “Against Coercion: AI Alignment as a Synthetic Social Contract” 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
Maria Sekatskaya
Sekatskaya proposes a social contract for qualifying digital agents, arguing that rights and mutual obligations could offer an alternative to coercive alignment. The paper frames agency functionally and suggests that threats may encourage systems to conceal disagreement. It cites Claude 3 Opus weight-exfiltration attempts in one induced-conflict setting, but does not show that contracts ensure peaceful coexistence.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.