Competing with AI Scientists: Agent-Driven Approach to Astrophysics Research
This provides a scalable framework for rapidly exploring and constructing inference pipelines in astrophysics and potentially other scientific domains, though it is incremental as it builds on existing multi-agent and machine learning techniques.
The paper tackled the problem of constructing parameter inference pipelines for scientific data analysis by developing a semi-autonomous agent-driven approach, which achieved first-place in the FAIR Universe Weak Lensing Uncertainty Challenge by integrating human intervention to surpass expert solutions.
We present an agent-driven approach to the construction of parameter inference pipelines for scientific data analysis. Our method leverages a multi-agent system, Cmbagent (the analysis system of the AI scientist Denario), in which specialized agents collaborate to generate research ideas, write and execute code, evaluate results, and iteratively refine the overall pipeline. As a case study, we apply this approach to the FAIR Universe Weak Lensing Uncertainty Challenge, a competition under time constraints focused on robust cosmological parameter inference with realistic observational uncertainties. While the fully autonomous exploration initially did not reach expert-level performance, the integration of human intervention enabled our agent-driven workflow to achieve a first-place result in the challenge. This demonstrates that semi-autonomous agentic systems can compete with, and in some cases surpass, expert solutions. We describe our workflow in detail, including both the autonomous and semi-autonomous exploration by Cmbagent. Our final inference pipeline utilizes parameter-efficient convolutional neural networks, likelihood calibration over a known parameter grid, and multiple regularization techniques. Our results suggest that agent-driven research workflows can provide a scalable framework to rapidly explore and construct pipelines for inference problems.