As the growth of renewable energy and rising threats from climate change increase the number and severity of power outages, artificial intelligence technology could help grid operators avoid blackouts.

Princeton researchers are exploring AI techniques that treat energy grid operation like a game that’s played against the environment — the better the grid stays up in response to fluxes or weather hazards, the greater the win.

“The nature of this game is very different than the kinds of things that have been successful before,” such as AI algorithms that play chess or Go, said computer science professor Ryan Adams. “The number of things that’s possible to do is enormous.” In addition, actions taken on a power grid have what researchers call continuous values. Rather than placing a game piece or flipping a switch, the actions involve tweaking output levels and distribution patterns.

Adams and his team are working with Siemens, an infrastructure technology company, developing solutions to modernize the energy grid, including renewables integration and autonomous systems.

Tackling real-world issues can move AI and computer science forward. Adams said that power grids are fertile ground for addressing problems that are relatively unexplored in AI research.