[image-caption title="NRECA%E2%80%99s%20Venkat%20Banunarayanan%2C%20Nick%20Pascale%20and%20Pat%20Mangan%20offer%20their%20insights%20on%20governance%20of%20artificial%20intelligence%20use%20at%20a%20March%2010%20PowerXchange%20breakout%20session.%20(Photo%20By%3A%20Jerry%20Mosemak%2FNRECA)" description="%20" image="%2Fnews%2FPublishingImages%2Fpx-ai-breakout-story.jpg" /]
ATLANTA—Electric cooperatives should create a policy, along with processes and safeguards, to govern their use of artificial intelligence, NRECA experts advised at PowerXchange.
Co-op leaders need to ask themselves: “Do we want to use AI? If we do, what’s the policy going to be?” said Nick Pascale, deputy general counsel for NRECA.
The role of co-op directors, he said, should be to raise the issue but leave the details of the policy to managers who know how employees are going to use AI on a day-to-day basis.
“If you don’t have a policy because you don’t think your employees are using AI, that’s probably not a good approach because they probably are,” Pascale said.
If a co-op decides to let employees use AI, there needs to be a way to enforce the policy that governs it, he advised.
“Without some enforcement, the policy is of no consequence,” Pascale said.
Co-ops and other utilities are looking to use AI to help them in key areas such as load forecasting, grid management and improving member services. It could be used, for example, to analyze data from sensors throughout a co-op’s system to detect anomalies and gain a deeper understanding of what’s happening on its grid.
“It’s important to understand what AI is being used for,” said Venkat Banunarayanan, NRECA’s vice president of integrated grid. “Are you using it to analyze data? Are you asking it to make specific decisions for you? What are the policies, processes and safeguards needed to ensure that it is accurate?”
Co-ops should never just “set it and forget it” when it comes to AI, he said.
“You can’t trust it blindly—never,” Banunarayanan said. “You’ve got to find out what data is going into it and what exactly the AI is doing to get you what you need.”
As a test of its performance, he suggested, do the same task manually and compare the results obtained with those from AI and ensure that any differences are explainable and make sense. If the differences in results are not understandable, AI could be producing bad results.
Pat Mangan, NRECA’s senior director of governance education, said AI must still be guided by human expertise.
“Keep a human in the loop,” he said. “Test the heck out of it before you set it loose, put some guardrails on it and continue to monitor it often.”
For more information, check out NRECA’s sample AI policy.