New Autonomous Science Roadmap: Verification Now Harder Than Discovery

The field of autonomous science has moved faster than anticipated since the original AISLE roadmap was released a year ago. Multi-agent systems are now producing experimentally validated hypotheses, self-driving laboratories have become more interoperable, and domain foundation models have raised the capability ceiling. The Genesis Mission has placed autonomous experimentation at the center of U.S. federal science strategy, with industry emerging as a primary actor. However, progress has met a sobering counter-current: a corrected flagship discovery result, benchmarks showing that agents which rival experts on closed-ended questions complete only a fraction of open-ended research, and fabricated citations surfacing at leading venues. The authors read this as the defining tension of the field: producing a candidate discovery is no longer the hard part, but verifying it is, and this asymmetry now limits autonomous science more than raw model capability.

The updated roadmap reorganizes around seven dimensions, elevating two former cross-cutting concerns—trust, verification, and reproducibility, and safety, security, and governance—to first-class status. It reassesses the original 14 milestones and adds four new ones (M15–M18). The path forward spans two years: the first year concentrates on interfaces, protocol adoption, and the scaffolding of verification; the second targets federation, zero-trust coordination, and governance. Throughout, the grassroots network is positioned as the interoperability fabric that lets national programs, international initiatives, and commercial platforms connect rather than re-silo. The roadmap provides a concrete structure for ensuring that autonomous science remains both capable and trustworthy.