Recently, with the powerful assistance of Claude, a large model from Anthropic, astrophysicists successfully created the first complete ultraviolet all-sky map in human history. In this vast star map containing 119 million stars, approximately one-third of the sky areas were completed by Claude based on previous multi-wavelength astronomical observation data. This achievement successfully filled historical gaps that had existed for the past 50 years due to reasons such as ultraviolet survey telescopes deliberately avoiding bright regions of the Milky Way.

In astronomical research, creating an all-sky map is a very demanding and complex systematic project. The tedious work that originally required astrophysicists to spend several weeks manually processing has seen a significant efficiency leap with the involvement of AI. Claude efficiently completed the entire workflow, including data collection, calibration, cross-calibration, and completion calculations, by flexibly scheduling multiple agents, shortening the original period of several weeks to just a few days.

What is more astonishing is that the areas completed by AI, after strict comparison, showed an error rate of only around 10% compared to real measured values. To ensure scientific rigor, the model also precisely marked the measured or calculated attributes of each pixel in the final image, along with the corresponding error range.

This breakthrough practice demonstrates that fundamental scientific research problems that have been long accumulated due to limited human resources are gradually being taken over by AI agents. In future frontier explorations, human researchers can step away from tedious low-level data processing and focus more on controlling the direction of research and identifying core scientific issues.