A look at how Americans spend their days reveals the questions they ask Google’s AI most often.
Some subjects break that pattern. People ask about government paperwork, health, money, legal issues, and what to buy far more often than they deal with any of those things. They don’t ask much about eating, getting dressed, cleaning, or watching TV, even though they spend most of their time there.
This number comes from Google’s AI & Economy ATLAS, a report published by researchers at Google and Google DeepMind. Google built ATLAS to examine how people use AI at work and at home, but says much of the household side is not represented in official economic statistics.
Covering 14.65 million interactions from Gemini apps, AI modes in search, and Gemini APIs. Google extracted U.S. non-work conversations from apps and AI modes and compared them to the U.S. Time Use Survey, which records what Americans do on a full day.
With AI modes included in our data, we now have a new understanding of what types of queries people bring to Google AI Search.
This is what I found.
When using AI to stay ahead of time
Google compared how often the topic came up in AI conversations and the amount of time people spent on it. Government services and citizen obligations show the largest difference, at about 20 to 1. This includes licenses, taxes, fines, and votes.
5 activities where AI conversations and everyday time diverge
Percentage of conversations in Gemini and AI modes outside of work in the US compared to percentage of active time outside of work in the US. Work and sleep are excluded. From April 6th to 19th, 2026.
timeshare gets bigger
1 x parity
AI conversations
bigger
Eating and drinking
Approximately 18 times
consumer purchases
almost 3 times
education
5.8×
Professional and personal care services
7 times more
Government services and citizen duties
almost 20 times
1/20×
1/5×
1×
5×
20×
logarithmic scale
Source: Google AI & Economy ATLAS v1.0. The value indicates how large the primary share is when comparing the share of AI conversations for each activity to the share of non-work active time.
In other words, these topics come up in AI conversations far more often than humans do.
Professional care services and personal care services have the next largest difference, with a difference of more than 7 to 1. This includes AI conversations about doctors, lawyers, banks, and salons. Education is nearly 6:1. About 3 to 1 people buy things.
Further down, Google’s data shows gaps in homework and research, managing your health, hobbies, shopping comparisons, financial services, hobby writing, electronics and tools, car repair, and more.
For things people do with their hands or in the same place, the gap moves in the opposite direction. Americans spend far more time in a day eating and drinking than people talk about in AI conversations. The gap is approximately 1:18. Things like travel, sports, and caring for family are also less likely to come up in AI conversations than the amount of time Americans spend on them.
TV and movies, laundry and getting dressed, cleaning the house, preparing meals, etc. are all at the bottom. That means people spend far more time on these topics than Google’s AI covers them.
None of this means people overlook other subjects. Time and conversation flow naturally no matter where you are, and discussions focus more on socializing and leisure than anything else. The gaps mentioned here are where this pattern is most noticeably interrupted.
why is this important
Health, money, legal, and shopping issues all rank high on the gap. In May, Google shared data that listed health, food, and travel as the top 10 themes in AI mode.
People are bringing these questions to Google’s AI, spending more time than expected on those questions. From this, you won’t know if there is anything that will send traffic to your website. This report covers conversations within Google’s own products, but does not include click data. Same gap as Merchant Center AI Query Pilot. Retailers were made aware of people’s questions without being told if they had sent someone somewhere.
Looking to the future
Google calls issues related to health care, law, money, and government high friction. About half of them came after hours, at night, early in the morning, and on weekends. The report fails to address what would have happened without Google’s AI. Licensing questions at 11 o’clock at night can become regular searches, wait until the morning, or get no response.
