AI tool generates realistic extreme weather scenarios for risk assessment

MIT engineers have developed a machine-learning algorithm that can produce plausible extreme weather events, such as intense rainstorms, without relying on historical records. The tool learns from regional weather data to exclude implausible scenarios and projects the size, intensity, and duration of rare events. This could help communities prepare for worst-case climate impacts.