Research program / tracks
Four permanent tracks
- T1
Collaborative Intelligence
When does dividing reasoning across specialists help, and when does it fragment evidence?
Communication protocols between agents, evidence exchange versus conclusion sharing, correlated errors, and how disagreement should change a conclusion.
- Rounds
- 4
- Hypotheses
- 1
- T2
Artificial Cognitive Architecture
How should capable models be organized into a system?
Role specialization, working memory, selective activation, controllers that allocate effort, and brain-inspired ideas treated as testable engineering hypotheses, not emulation.
- Rounds
- 25
- Hypotheses
- 16
- T3
Reasoning & Verification
How does a system know when an answer is ready to release?
Claim-level evidence review, bounded repair, revision drift, terminal release gates, and calibration of the final revised answer.
- Rounds
- 22
- Hypotheses
- 86
- T4
Evaluation Science
How do we measure whether an architecture change actually improved reasoning?
Endpoint definition, scorer failure modes, matched-cost comparisons, ablations, and the separation of diagnostic scores from adjudicated quality.
- Rounds
- 23
- Hypotheses
- 17