Health AI's environmental footprint needs equal attention, paper argues
A new paper in Artificial Intelligence in Medicine examines the environmental costs of using AI in health care. Co-author Alok Mishra of NTNU argues that energy-hungry models and data centers consume large amounts of electricity and fresh water. The work calls for environmental impact to be considered alongside privacy, bias, fairness, and transparency in AI development.
A new paper in Artificial Intelligence in Medicine examines environmentally sustainable AI in health care. Its co-author, NTNU professor Alok Mishra, has argued for a broader “Green AI” approach that weighs energy use, data-center electricity demand, and freshwater consumption alongside clinical benefits.
The International Energy Agency projects data-center electricity use may double by 2030, matching Japan’s national consumption. The agency also notes AI could improve grid operations and save up to 175 gigawatts of transmission capacity, enough to power Oslo for a year. Health care remains an eager AI adopter, using it for diagnostics, digital clinics, and faster analysis of patient information.
If medical AI’s environmental costs go unmanaged, patients and health systems may face indirect consequences through rising energy demand, water use, and emissions, while clinicians could still gain efficiency. Communities near data centers may also be affected. Treating environmental impact alongside privacy, bias, fairness, and transparency could shape how hospitals, developers, and regulators assess AI tools, though trade-offs may remain.