Data reveals that while most American adults have never used AI to perform an indicated task, they are more likely to use AI for humanities-related tasks. Of the 12 tasks included in the survey, Americans were most likely to have used AI to learn about a subject’s history, with 46% having used AI for this purpose at least to some extent and 19% reporting using AI daily or weekly to accomplish this task. A significant proportion of people used AI to revise their texts (44% revise and 15% do so at least weekly). A similar proportion used AI to create paragraphs or more of text (40% at least once and 15% at least weekly).
Among other humanitarian activities, a significant proportion of people at least sometimes used AI to translate texts from another language into English (37%) or to summarize long texts or books (34%). A similar percentage of Americans used AI to perform other types of tasks, such as analyzing data (38%), generating images (36%), solving math problems (31%), and brainstorming creative ideas (31%). A much smaller proportion used AI to create poems or songs (16%) or develop computer programs (14%). Among humanities-related tasks, Americans were least likely to have used AI to evaluate ethical choices (23%).
For most tasks asked in the survey, younger Americans were more likely to use AI on a daily or weekly basis, and those aged 30 to 44 in particular were significantly more likely than those 60 and older to regularly use AI for text creation or data analysis. Conversely, for all tasks, older Americans (age 60 and older) were the least likely to say they had never used AI. About 60% said they would not use AI to learn the history of a subject. For all other tasks, more than 70% have never used AI.
For most tasks, higher-income Americans were more likely to use AI. People with an annual income of $100,000 or more were significantly more likely to use AI to write, revise, and summarize their texts than less wealthy Americans.
The use of AI in humanities tasks is also positively correlated with education level, and this association is not found in other types of tasks. Americans with a bachelor’s degree were about twice as likely to use AI to write, revise, and summarize their writing daily or weekly compared to Americans without a degree. 23% of people with a bachelor’s degree report relying on AI daily or weekly to revise or improve what they’ve written, compared to just 10% of those with a college or associate’s degree and 12% of those with a high school diploma or less. 18% of bachelor’s degree holders reported using AI to summarize text, compared to less than 10% of degree holders. Conversely, Americans who graduated from high school were more likely than those with higher education to rely on AI to solve math problems on a daily or weekly basis (13%, compared to 7% among Americans with at least some college degree).
For almost all tasks, there was little difference in the proportion of men and women who used AI on a daily or weekly basis. However, more men (23%) than women (15%) report using AI to find out the history of something.
Other recent research points to widespread public concern about the impact of AI use, yet its use in the workplace is on the rise. In this study, only a small percentage of the population (1-6% depending on the scale used to measure intensity) were high-intensity users. (Typically someone who used AI for various tasks on a daily or weekly basis). High-intensity users employed AI for significantly more tasks than low-intensity users. On average, high-intensity users used AI at least weekly for at least nine of the 12 tasks examined in the study. Those who used AI less intensively (usually only monthly or less than monthly) used AI for only two tasks. Those who reported rarely using AI used it for the fewest tasks.
Although many questions remain unanswered, this study provides some perspective on the extent of the use of AI to explore the humanities compared to other common forms of intellectual activity. If you have thoughts or suggestions about how this research can be improved or expanded in the future, please contact rtownsend@amacad.org.
