The Interplay Between AI and Power Systems

Energy consumption from the use of generative AI and crypto mining have increased substantially over the last couple of years, and these computing demands present impacts on energy system reliability, market prices, and carbon emissions.

In his talk, Le Xie from the Power and AI initiative (PAI) at the Harvard School of Engineering and Applied Sciences dives into the challenges facing the U.S. power grid, including aging infrastructure, the surging energy demands of AI data centers, and blockchain mining. The talk focuses on key research questions centered around development of foundational models for power grids, use of AI algorithms for grid operations, and the future of the energy grid to support advances in the use of AI. Additionally, Le discusses regulatory and infrastructure challenges, aiming to accelerate grid modernization and data center deployment.

The talk also introduces PowerAgent, an open-source agentic AI designed to support various workflows within the power systems domain. A demo video showcases the use of PowerAgent by providing an objective, scenarios to analyze, and recommendations to mitigate identified issues. PowerAgent is composed of three open-source repositories such as PowerFM (foundational models), PowerMCP (collection of MCP servers for power system software), and PowerWorkflow (collection of agentic workflows for power system applications).

PowerAgent GitHub repository: https://power-agent.github.io