Published on
Millions of people are using artificial intelligence (AI) agents for learning or productivity in their personal lives, in what researchers say is the first study on their adoption.
AI agents are like online assistants that can plan and execute complex tasks with little human supervision, based on a user’s request. In 2025, many of the world’s biggest AI companies, including Amazon, Google, Microsoft, and OpenAI launched or expanded their own digital assistants.
A Harvard University researcher teamed up with one such company, Perplexity AI, to examine data from the startup’s AI browser and digital assistant, Comet, which launched in July 2025.
The researchers analysed hundreds of millions of queries to understand how the agent was being used and published their findings, which have not yet been peer-reviewed, online this week.
The researchers classified users based on their jobs and the ways they typically used the agent.
People who started using AI agents early on, as well as users from wealthier, more highly educated countries were more likely to “adopt or actively use the agent,” the researchers said.
More than 70 per cent worked in a digital or knowledge-intensive field, for example academia, finance, marketing, or entrepreneurship, the study found.
The fields with the fewest AI agent users were those that “require interacting with the physical environment,” such as energy and agriculture, it said.
Thirty-six percent of all tasks assigned to an AI agent were considered “productivity and workflow” tasks such as creating or editing documents, filtering emails, summarising investment information, or creating calendar events.
The second most common tasks were related to “learning and research,” with 21 percent of queries asking an agent to summarise course materials or research information.
Other popular tasks included assistance with shopping, travel, and job-related searches.
The users asked their AI agents for more help in their personal lives than their professional ones: 55 per cent of questions were related to their after-hours lives compared to 30 per cent that were related to work.
Another 16 per cent of queries were related to education.
The study showed how people used the AI agent evolved over time. Users who started with simple, personal tasks involving topics like travel and media often pivoted over time to more labour-intensive queries that had to do with productivity, learning, and careers.