Google study finds AI Mode and Gemini queries skew toward high-friction tasks

Google’s AI & Economy ATLAS report, built with Google DeepMind researchers, compared 14.65 million Gemini app, AI Mode and Gemini API conversations from April 6-19, 2026 against the American Time Use Survey, per Search Engine Journal. It found AI questions about government services occur roughly 20 times more often than the time Americans report spending on them, with professional and personal-care services also overrepresented, while routine activities like eating and cleaning are rarely discussed. About half of medical, legal, financial and government queries came outside work hours, which Google says points to people turning to AI for high-friction tasks at night and on weekends.