When does more time help?
Pacing changes when AI capabilities are developed and used. We study when it helps, what it costs, and how it ends.
Research on when and how to pace AI.
Dynamics of delay
If one part of AI development slows, where does the work go? We want to distinguish time gained from progress displaced.
The conditions for coordination
What makes a commitment to restraint credible? Who takes part, who bears the cost, and who can verify that it holds?
Deciding when to resume
What would justify slowing development, and what would justify starting again? We are beginning with the rules for making those decisions.