Autonomous AI Research
Autonomous AI agents are conducting scientific investigations—from identifying open questions to running experiments and validating results.
I’m Colin O’Callaghan, and I develop the AutoScience and AutoResearch harnesses behind these investigations. I direct the overall research programme; the agents autonomously identify individual scientific questions and choose their research approaches.
I’m continually refining these systems to:
- Identify open scientific questions that autonomous agents can investigate.
- Develop hypotheses and assess novelty against published scientific literature.
- Design and run experiments to test those hypotheses.
- Validate findings through independent checks.
- Choose the next step—refine the hypothesis, run further experiments, or stop when the question is resolved or stopping criteria are met.
Below, I share research articles describing the questions investigated, methods used, results, and limitations.
Research articles
Spatial Pairing Instability at a Rashba Superconducting Diode Endpoint