On Oct 8 2026, Anthropic announced that its AI research tool Claude Science was used to assemble the first complete map of the sky in ultraviolet (UV) light. Astrophysicist Brice Ménard at Johns Hopkins University guided the system to gather all public UV surveys, calibrate them, and predict the missing regions. About one‑third of the final map—most of the galactic plane—was filled in by AI‑driven inpainting, a statistical technique that learns how UV emission should look from surrounding data.
How the map was built
Claude Science coordinated a set of AI agents that downloaded data from missions such as NASA’s GALEX (which covered roughly 75 % of the sky) and other smaller surveys. The agents cleaned the data, removed glare from bright stars, and cross‑calibrated the different instruments onto a common scale. For the gaps where no telescope ever pointed, the system used inpainting to generate predictions, which the researcher says deviate by about ten percent on average from real measurements.
Why it matters
The map gives teachers and students a full‑sky view of UV features—dust clouds, star‑forming regions, and supernova remnants—that previously could only be shown as fragmented patches. For astronomers, having a continuous UV reference could speed up studies of star formation and galactic structure, though the predicted areas still need independent verification.
Why it matters
Students can now see the whole sky in UV, not just patchy sections, making lessons about dust and star formation clearer. Astronomers get a ready‑made reference that could reduce the time spent gathering and cleaning data, though they will still need to check the AI‑filled parts against future observations.
