Reproduce “Living constitutions: Participatory alignment for sovereign foundation models” 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

Stine Lyngsø Beltoft, Jonas Vistrup, Peter Schneider-Kamp, Lukas Galke Poech

The authors propose an AI constitution that records agreement, persistent disagreement, and open issues. They describe its deliberative process and test robustness in an agent-based simulation. With a 50-seat committee split equally between citizens and uncorrupted developers, citizen bad-actor shares up to 0.55, 0.65, and 0.80 met a 95% pass rate for capture, mixed attacks, and dilution, respectively; results remain conditional on model assumptions.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.