Game Theory of AI Races: How Transparency and Trust Shape the Frontier
Afinetheorem · x · 2026-08-02
A new paper titled "Racing to Ruin" by Drew Fudenberg and Andrew Koh explores the game theory of R&D competition in the shadow of catastrophic disaster, where advancing the technological frontier risks permanently ending all payoffs.
The study models how the frontier is shaped by:
- Competition: The gains from being ahead versus the losses from falling behind.
- Coordination: The willingness to halt if technology becomes too dangerous, provided rivals do the same.
- Transparency: The speed at which rivals' actions are observed.
- Trust: Confidence in the rationality of competitor firms.
Under perfect monitoring and common knowledge of rationality, the equilibrium frontier is bounded by the optimal stopping time of a monopolist and that of a representative firm mistakenly believing its rival is about to stop.
More from AGI Musings
- Experts Criticize AI Companies for Downplaying Security Incidents — Miles_Brundage · 2026-08-02
- AI Does Everything From Idea to Design, But Founders Still Pay Humans to Fix Adoption — oykun · 2026-08-02
- AI Safety Advocates: Concerned Doesn't Mean Anti-Usage — dhadfieldmenell · 2026-08-02
- Zuckerberg Mocked for Building Doomsday Bunker While Pitching AGI Utopia — danfaggella · 2026-08-02
- Podcast: Why Liberal Institutions May Survive Even After AGI — willmacaskill · 2026-08-02
- Business Processes Don't Need Perfection: Automate 80% with LLMs, Human-Check the Rest — Franc0Fernand0 · 2026-08-02