AI Is Changing Creative Work, but the Arts Aren't Disappearing

Generative AI has intensified concerns about job loss in creative fields. If software can produce images, music, and text in seconds, it is easy to assume that artists and other creative professionals will be among the first workers displaced.

But early evidence suggests the story is more complicated.

A recent study in the Journal of Cultural Economics, drawing on the Gallup Panel workforce studies and federal labor market data, finds little evidence so far that generative AI has broadly reduced artists’ earnings. Across multiple national datasets, artistic occupations that are more exposed to large language models have not seen the sharp wage declines many expected.

What appears to be changing instead is how creative work is organized.

To understand where generative AI might matter most, the analysis uses an occupational exposure to generative AI index published in 2024 that estimates the share of tasks in an occupation that large language models could plausibly perform or assist with.

The results show wide variation even within the arts. Some occupations contain a large share of tasks that overlap with the capabilities of generative AI systems. Music directors and composers, for example, have an exposure score of about 0.70, meaning a substantial portion of their tasks involve composition, arrangement or other forms of structured creative production that AI tools can help draft or modify. Special effects artists and animators follow with exposure around 0.54, while disc jockeys, art directors, and producers and directors cluster around 0.50.

Other artistic occupations are far less exposed. Dancers, whose work is grounded in physical performance and embodied movement, have an exposure score near 0.04. Actors are around 0.18, while craft artists and choreographers fall around 0.27 to 0.28. In these fields, the core of the work involves live presence, interpretation and physical skill that generative systems cannot easily substitute. This variation makes it possible to examine whether occupations that are more exposed to AI actually show worse labor market outcomes.

The evidence does not show large negative effects when examining the impact of AI on jobs. Using employment and wage statistics from the Bureau of Labor Statistics between 2017 and 2024, earnings trends for artistic occupations with higher exposure to generative AI look broadly similar to those with lower exposure. The estimates are slightly positive, though they are not statistically distinguishable from zero.

Employment patterns are more mixed. In 2023, some highly exposed artistic occupations experienced weaker employment growth relative to less exposed ones.  Even so, the differences are modest and far from the widespread job losses that discussions of AI and job displacement often assume.

Worker-level data offer another perspective on the impact of AI art on artists. In the American Community Survey by the Census Bureau, artists in more exposed occupations show a modest increase in earnings in 2023 that fades somewhat in 2024. At the same time, total hours worked rise more clearly beginning in 2022 and remain elevated through 2024. The Census Bureau data are preferable here because many artists are freelancers rather than salaried workers, which is what the Bureau of Labor Statistics data capture. Examining workers directly produces a more complete picture.

To understand why effects might be positive, the analysis draws on Gallup Panel workforce studies. Employees in artistic occupations report somewhat higher AI use than the workforce overall. Among occupation-defined artists, roughly one in four say they use AI frequently, compared with about one in five workers across the broader economy. Equally as important is where AI appears inside the creative process.

Artists are more likely than other workers to report using AI for idea generation and creative exploration. They also report using it to automate small tasks, consolidate information and support collaboration. Artists are, not surprisingly, less likely to use AI for operational tasks such as customer interaction or equipment management.

These patterns suggest generative AI playing a role primarily in the early stages of creative work — helping artists experiment with ideas, iterate quickly and organize parts of the creative workflow. Generative AI could also enable artists to have more agency over their own careers by augmenting their ability to produce branding documents, craft outreach, and automate otherwise mundane tasks with travel and accommodation.

Creative industries, especially classical music and the performing arts, have navigated technological disruptions before. The phonograph and early recording technologies raised fears that recorded music would replace live performance, yet recordings ultimately created new markets while concerts remained central to artistic life.

Similar anxieties arose when photography reduced demand for portrait painting and when digital compression disrupted recorded music revenues. In each case, technology changed how people produced and distributed creative work, but the arts adapted rather than disappeared.

Generative AI appears to be producing another period of adjustment. AI in creative industries is altering tasks and workflows. But early data do not show a collapse in artistic work. They show an industry beginning to reorganize around a new technology.

That experience may also offer lessons for broader questions about AI and the future of work, particularly in knowledge-intensive fields where generative AI adoption is accelerating. In industries such as marketing, communications and design — where work similarly combines creativity, iteration and judgment — the technology may reshape how people perform tasks without eliminating the underlying roles, and may create new ones in the process.