Grid operators don’t have to fear losing their control room jobs to artificial intelligence — at least not yet.
The Electric Power Research Institute released results of a survey Tuesday that quizzed the leading AI large language model apps on questions that electrical engineers, planners and other grid professionals should know.
OpenAI’s GPT-5 got the best score, followed by Google’s Gemini 2.5 Pro and Anthropic’s Claude Sonnet. Perplexity’s Sonar Pro was fourth, and xAI’s Grok 4 came in last. But they didn’t go to the head of the class.
While GPT-5 and Gemini managed B-minus grades on the easiest questions helped by multiple-choice prompts, and the others trailed, EPRI reported that all the AI assistants embarrassed themselves on the hardest, open-ended questions. GPT-5 scored 63 percent right. The rest were between 46 percent and 59 percent correct, exposing reliability risks, EPRI said.
