Anthropic's Claude Science has successfully created the first complete ultraviolet map of the entire sky. The project demonstrates how artificial intelligence can handle tedious data tasks that researchers often lack the time to complete, as detailed by Johns Hopkins astrophysicist Brice Ménard on the Anthropic website.
Ultraviolet light is crucial for revealing dust illuminated by starlight, including clouds around young stars and rings left by stellar explosions. Previously, a complete UV map did not exist because Earth's ozone layer blocks ultraviolet light, requiring measurements to be taken from space. While NASA's GALEX mission previously mapped about two-thirds of the sky, it notably skipped bright star-forming regions.
To achieve this, Claude Science coordinated AI agents that downloaded data from multiple space missions, calibrated it, and merged the information into a single dataset. The system filled missing areas using inpainting, a technique where the model learns from existing data to reconstruct gaps with predictions that averaged about ten percent deviation from actual measurements. The resulting map is intended to serve as teaching material, and Ménard suggests that AI could enable scientists to finally tackle similar delayed projects.
This achievement matters because it proves AI can automate complex data aggregation and fill critical observational gaps in scientific research.