Workplace Adoption of Generative AI

Generative artificial intelligence (AI) has recently emerged as a potentially transformative workplace technology. The ultimate impact of generative AI on the economy will depend on how many workers adopt the technology, how intensively they use it, and for which tasks. In The Rapid Adoption of Generative AI (NBER Working Paper 32966), researchers Alexander Bick, Adam Blandin, and David J. Deming report on a nationally representative US survey of generative AI adoption at work and at home.

Using the Real-Time Population Survey, which mirrors the methodology of the Current Population Survey, the researchers surveyed over 5,000 Americans aged 18–64 in August 2024. They defined generative AI as “a type of artificial intelligence that creates text, images, audio, or video in response to prompts,” citing examples like ChatGPT, Google Gemini, and Midjourney.

The researchers found that 39.4 percent of respondents had used generative AI. Among employed respondents, 28 percent reported using generative AI for their job, with 24.2 percent using it at least one day in the previous week, and 10.6 percent using it every workday in the previous week. Outside of work, 32.7 percent used generative AI, with 25.9 percent using it at least one day in the previous week and 6.4 percent using it every day in the previous week. The most commonly used product was ChatGPT (28 percent) followed by Google Gemini (16 percent).

How do these adoption rates compare with those of previous technological advances? Two years after the release of ChatGPT, the first mass-market generative AI product, overall adoption rates were nearly double those of personal computers (PCs) three years after the release of the IBM PC in 1981. They are also higher than internet adoption rates at a similar stage. However, workplace adoption rates for AI (28 percent in two years) are similar to those for PCs (25 percent in three years), highlighting at-home use as a key contributor to faster overall adoption.

Men were 9 percentage points more likely than women to use generative AI at work, reversing trends of early PC adoption. Workplace usage declined with age, from about 34 percent for workers under 40 to 17 percent for those 50 and older. Educational attainment was correlated with adoption, as about 40 percent of workers with a bachelor’s degree or higher were adopters who used generative AI at work, compared to 20 percent of those without a college degree. Adoption was highest in computer/mathematical (49.6 percent) and management (49 percent) occupations, but even 22.1 percent of workers in blue-collar jobs reported using generative AI at work.

How intensively is generative AI used, and for what tasks? On days that employed respondents used generative AI for work, 23 percent used it for less than 15 minutes, 52 percent used it for between 15 minutes and one hour, and 25 percent used it for more than an hour. The most common applications of generative AI at work were to help with writing, searching for information, and obtaining detailed instructions.

— Leonardo Vasquez

The survey was funded in part by the Walmart Foundation, grant 252882.