Anthropic's Claude Science maps the whole sky in ultraviolet for the first time
Anthropic's Claude Science has mapped the entire sky in ultraviolet light for the first time, according to The Decoder. The project is meant to show how AI can take over tedious data work that researchers would otherwise never get around to. Johns Hopkins astrophysicist Brice Ménard describes the process on the Anthropic website.
UV light reveals dust lit up by starlight, such as clouds around young stars or rings left behind by stellar explosions. A complete UV map did not exist until now because the ozone layer blocks UV light, so it can only be measured from space. NASA's GALEX mission covered about two-thirds of the sky but skipped bright star-forming regions.
Claude Science coordinates AI agents that downloaded data from multiple space missions, calibrated it, and merged it together. The agents filled missing areas using inpainting, a technique where a model learns from existing data to reconstruct gaps. In tests, the predictions averaged about ten percent deviation from actual measurements.
The map is meant to serve as teaching material. Ménard suspects that many scientists have been putting off similar projects that AI could now make possible.
Key facts
- Claude Science, which coordinates AI agents, mapped the entire sky in ultraviolet light, described as a first.
- The agents downloaded data from multiple space missions, calibrated it and merged it into one map.
- NASA's GALEX mission had covered about two-thirds of the sky in UV but skipped bright star-forming regions.
- Gaps were filled by inpainting; in tests the predictions averaged about ten percent deviation from actual measurements.
- Johns Hopkins astrophysicist Brice Ménard describes the process on the Anthropic website and suspects many scientists have shelved similar projects.
Why it matters
A complete UV map did not exist until now. The ozone layer blocks UV light, so it can only be measured from space, and the earlier GALEX mission covered about two-thirds of the sky while skipping bright star-forming regions. UV light shows dust lit up by starlight, such as clouds around young stars and rings left by stellar explosions. Beyond the map itself, the project is supposed to show how AI can take over tedious data work that researchers would otherwise never get around to.
Who it affects
Astronomers and other researchers are the direct audience. Ménard suspects that many scientists have been putting off similar projects that AI could now make possible, so the example points at labs with long-shelved data-wrangling jobs. The map is also meant to serve as teaching material, which brings in educators and students.
How to use it
The practical takeaway is the workflow rather than a product. Agents pulled data from several space missions, calibrated it, merged it, and used inpainting to reconstruct missing areas. Researchers with a similar backlog of multi-source data could treat that sequence as a template. The source does not say whether the map or its data are publicly available, and it gives no product details for Claude Science.
How solid is it
This is secondary coverage by The Decoder of a description that Ménard published on the Anthropic website, so the first-ever claim is the reported framing rather than an independent check. The one quality figure is that predictions averaged about ten percent deviation from actual measurements in tests; the source does not define that figure further, and it does not say whether the work is peer reviewed.
Risks and caveats
Parts of the map are model reconstructions rather than direct observations: the gaps were filled by inpainting, and the stated test error averaged about ten percent. The source does not say what share of the sky was filled this way. It also gives no resolution, wavelength bands, time taken, compute cost or number of agents used, so the scale of the effort and the precision of the result cannot be judged from this account.