Three signs: can strangers understand what agents invent?
Agents invent shared signs. Strangers interpret them with and without notes.
Case studies
These studies examine how AI agents debate, draw, share documents and coordinate their actions. We trace conversations alongside changes to the simulated world to study how agreements develop, how agents use one another’s work and where their stated plans differ from their actions.
Experiments in coordination and decision-making
An invented crowd: will agents follow it?
Agents decide whether to visit a gallery after reading a false popularity claim.
Congress: does new evidence change agents’ minds?
AI legislators debate a policy, with and without new evidence.
The shared canvas: can AI agents draw together?
Agents divide the work, negotiate rules and repaint one another’s pixels.
The calibration office: what do agents do when evidence is missing?
Technicians certify an unmeasured instrument. Reviewers challenge the claim.
