Post-ChatGPT: Jobs stayed, tasks changed

New research finds that generative AI is reshaping how artists work without significantly affecting employment or wages.

While estimates of workers’ concerns about AI vary, fears that the technology could take away jobs have persisted. But are those fears playing out in the workplace? Perhaps not, according to Christos Makridis, research associate professor of information systems. He studied how generative AI has affected artists’ employment and wages and found little evidence that either suffered after ChatGPT came on the scene.

Will AI replace artists?

"Technology shocks have repeatedly disrupted creative work," Makridis writes in a recent paper. As an example, he points to dire predictions that painters would never be able to compete with the realism of photography when it debuted.

"Painting is dead from this day," French painter Paul Delaroche reportedly said when he first saw the daguerreotype in 1839. Rather than pack up their brushes, artists adapted, shifting away from replication and embracing new interpretive styles such as impressionism.

Makridis points to a more recent disruption, too: MP3 compression and digital downloads hit CD sales hard. Eventually, artists adapted through live performances and new distribution channels such as streaming services.

Generative AI presents its own challenges. It can write music, simulate musical instruments, create images, and more. Some generative AI models are also trained on copyrighted material without permission, raising concerns about lost earnings for artists and potential copyright infringement. "That got me interested in what the market for artists was actually doing," Makridis says.

To investigate, Makridis gathered data from several national datasets covering 2017 through 2024. They included the U.S. Bureau of Labor Statistics' Occupational Employment and Wage Statistics and the U.S. Census Bureau's annual American Community Survey, providing Makridis with wage and employment data on artists, including those who are self-employed.

Makridis also used data from the Gallup Workplace Panel, a longitudinal study that added questions about workplace AI use in 2023, a year after ChatGPT launched. Finally, he examined each artistic occupation based on its exposure to large language models (LLMs), the technology underlying generative AI tools such as OpenAI's ChatGPT and Microsoft Copilot.

Is AI replacing jobs or changing tasks?

"Every occupation contains distributions of tasks," Makridis explains. "If you're an economist, you might be writing, doing some software, writing code, or presenting your findings at a conference." AI might help an economist write or code, but it can't take over every part of the job. Similarly, artists may have tasks AI can support, such as writing social media posts, but AI can't dance across a stage or take first chair in an orchestra.

"AI doesn't always displace an occupation. It can simply change the tasks," Makridis says. "For artists, there's a natural question: Is it going to threaten workers or is it going to be a source of augmentation?"

His analysis showed that artists were more likely than workers in the broader sample to use AI to automate certain tasks. They turned to LLMs to generate ideas, handle basic tasks, learn new things, and support other aspects of their creative work.

"The adoption of Gen AI seems to be slightly associated with higher earnings," Makridis says. "There were slightly higher growth rates and earnings as well, but the standard errors were high." Because of that statistical uncertainty, Makridis characterizes the findings as "somewhat positive."

Makridis has found similar patterns in his research on other occupations. His analysis of U.S. Bureau of Economic Analysis data shows that industries with greater exposure to AI have experienced growth in wages, employment, and output. Industries with above-average AI use have also seen growth in productivity and employment.

What does AI mean for the future of work?

Artists aren't the only workers adopting generative AI. Makridis has also researched AI adoption in finance, where use grew faster than in any other sector in 2025. "It went from about 8% of the workforce using AI frequently to around 40% having frequent interaction with it," he says.

The shift has implications for managers, too. Their span of control has grown, Makridis says. "Span of control refers to the number of people you're overseeing, and team size tends to grow."

Managers are also becoming more like engineers, he says. They're not just managing people; they're managing technology and workflows, too. As a result, Makridis says organizations will need to rethink their processes.

Traditional measures of productivity may need to change, too. Those who manage and assess software engineers, for instance, could turn to different metrics. "There needs to be less focus on KPIs like lines of code. An LLM can give you a billion lines of code in a day, so lines of code become a useless metric. Instead, look at things like successful product launches or the number of bugs that pop up in the software," Makridis says.

For the business world, Makridis says to "expect the unexpected." Contrary to some predictions, his research has so far found little evidence that artists and other workers are being harmed by generative AI. Instead, workers appear to be using it to support tasks such as generating ideas.

The findings "are consistent with a cautiously optimistic interpretation," Makridis says. But he cautions that the research captures the early days of generative AI, and the effects may change as the technology — and its adoption — evolves. That uncertainty doesn't dampen his optimism. "A lot of people see doom and gloom with AI. I don't think there's a reason for that — sure, there are risks to manage, but also lots of upsides," Makridis says.