IBM is launching a new open source AI model to get NASA back to the Moon — and making petabytes of lunar data available to study (www.techradar.com)

IBM and NASA have unveiled an open source AI model designed to enhance lunar research as part of efforts to return humans to the Moon. The NASA-IBM Lunar Foundation model, available on Hugging Face, enables researchers to analyze over five decades of lunar observation data more efficiently. It is expected to significantly improve the identification of geological features, such as potentially hazardous ice deposits and craters, which are crucial for building a sustained human presence on the Moon. This initiative is particularly significant for the AI/ML community as it addresses the limitations of current methods, which often involve labor-intensive manual analysis or task-specific AI models that can be inaccurate and computationally expensive. By providing a singular foundation model adaptable for multiple geological investigations, the tool reduces errors and streamlines research processes. The release of a comprehensive dataset—including 30 spatially aligned layers from various missions—marks a milestone for the community, offering unprecedented access to machine learning-ready lunar data. The collaborative efforts are positioned to foster advancements in lunar exploration while enhancing our understanding of the Moon’s geological history and potential resources.