Keep up with the innovative tech transforming business
Tech Brew keeps business leaders up-to-date on the latest innovations, automation advances, policy shifts, and more, so they can make informed decisions about tech.
Running AI might eat up massive amounts of energy, but can the technology also help the electrical grid run more smoothly?
That was one of the questions posed in a pair of Department of Energy reports this week that outlined ways AI might be used to bolster the country’s power infrastructure. Those include smarter distribution of electricity through better stress predictions, more thorough renewable energy forecasts, and the use of large language models (LLMs) to speed up permitting.
The reports were produced as part of a sweeping executive order President Biden signed around AI six months ago. Part of its mandate was to direct federal departments and agencies to survey the risks and potentials of the technology.
Impending stresses: Experts expect factors like erratic weather patterns, electrification pushes, and vast new data centers—many of which power AI itself—to significantly strain the US’s aging electrical grid in coming years.
But the reports said that more comprehensive data about usage and AI-powered analysis might also help better route and mete out all this power. The authors discuss AI use cases like predicting maintenance needs, detecting anomalies, and forecasting weather and energy prices to make renewables as efficient as possible.
“Emerging applications for AI offer the potential to enable change on the grid at a nonlinear pace and scale,” one of the reports, titled AI for Energy, said.
As for the new wave of LLMs, the reports discuss how they might be fine-tuned on the jargon of permitting applications to do things like determine characteristics of a project that might lead to a longer review. The report also discusses using LLMs to better digest long lists of public comments tacked onto a project.
Risky business: All that said, the reports also point to a number of risks that might crop up if AI is “used or deployed naïvely” on a grid that the AI for Energy authors describe as “among the most complex machines on Earth.”
In this case, risks of AI applications span everything from drift—when a model operates differently from its intended purpose—to the electrical usage of the AI itself, and added cybersecurity vulnerabilities.
“Careful consideration of how AI deployment affects different stakeholders and industries can mitigate downstream risks or unforeseen hazards,” the AI for Energy report said. “The Department of Energy is committed to the safe, secure, and responsible deployment of AI for the clean energy economy.”