Nature Communications · 13 October 2026 · From issue 51
Reinforcement learning guided robotic platform that investigates and discovers asymmetric reactions
In lab tests, a robot using trial-and-error learning found four new mirror-shaped chemical reactions on its own.
Patryk Klochowicz, Ahmad F. Kassir et al.
A robot chemist worked for 30 days without a human guiding each step. It made over 3,000 decisions on its own and found four new chemical reactions. These reactions build molecules with a special mirror-image shape. That shape matters a lot in medicine design.
The number of choices the robot made while searching for new mirror-shaped reactions.
Lab-only test with one acid helper, not yet broadly proven.
Integrity screen: passed (3 checks) Checked 14 October 2026. Retraction record: none. DOI resolves at doi.org. Metadata record found (Nature Communications). Read the source
This screen checks process signals. It does not tell you the findings are true.