Artificial intelligence has the potential to significantly reshape Michigan’s economy, but its effects will depend heavily on how quickly businesses adopt the technology and how policymakers address issues of ownership, competition and regulation, University of Michigan economist Justin Wolfers told lawmakers late last week.
Wolfers, a professor of public policy and economics at the University of Michigan, testified during a Michigan House Appropriations Committee hearing focused on artificial intelligence and long-term economic growth. He described AI as a “white-collar revolution” that primarily affects cognitive and office-based work rather than traditional manual labor.
Drawing on recent research, Wolfers said early evidence suggests AI can substantially increase worker productivity.
In studies cited during his testimony, access to AI tools reduced the time needed to complete tasks by 14% to more than 50%, while also improving the quality of work produced. He noted that these gains often benefit lower-performing workers more than higher-performing ones, potentially reducing inequality within workplaces.
“Typical productivity growth in the U.S. economy might be about 1% in a year,” Wolfers said, adding that AI-driven gains in specific tasks can represent “a generation’s worth of productivity growth.”
Wolfers cautioned, however, that widespread economic transformation is unlikely to happen overnight. He said history shows major technological revolutions — including steam power, electrification and personal computers — took decades to translate into sustained productivity gains because businesses had to reorganize how work was done.
Current AI adoption remains limited, Wolfers said, with surveys suggesting about 10% of U.S. businesses have meaningfully adopted AI tools, while another 14% plan to do so within six months. Michigan’s adoption rates appear broadly similar to national trends.
Because of that slow uptake, Wolfers said AI’s near-term effect on gross domestic product may be modest. Under conservative assumptions, AI could raise GDP by about 1% over the next decade. More optimistic projections from financial institutions and economists place potential GDP growth closer to 8% to 12% over that same period, depending on cost reductions, investment in data centers and broader organizational changes.
The most significant long-term impact, Wolfers said, could come if AI substantially boosts innovation itself.
“If AI raises the productivity of innovators,” he told lawmakers, “it’s like getting more research and development for free.” Over time, he said, that could compound into much higher economic output, potentially raising GDP by 30% by mid-century.
Wolfers also warned that AI’s benefits will not be evenly distributed without careful policy choices. He emphasized that economic outcomes depend less on the technology itself than on who controls it and how markets are structured.
Using a series of hypothetical examples, Wolfers explained that AI could either improve workers’ quality of life or concentrate wealth among employers or technology firms, depending on ownership and competition. He said strong competition among AI providers helps keep costs low, while monopolies — whether among AI companies or upstream suppliers such as chip manufacturers — could capture most of the economic gains.
“Small changes in regulation could have enormous effects on how this plays out,” Wolfers said.
He highlighted potential policy challenges related to energy demand, data center construction, semiconductor supply and the pace of regulatory decision-making. Bottlenecks in any of those areas, he said, could slow adoption or concentrate economic power.
Wolfers said AI is especially likely to affect professional services, including law, healthcare administration and knowledge-based roles. In healthcare, he cited growing use of AI for documentation and diagnostics, noting that studies show AI systems can match or exceed human performance in some tasks, though public trust remains a barrier.
Manufacturing and healthcare — two pillars of Michigan’s economy — are both likely to see significant change, Wolfers said, particularly in roles that combine technical skill with decision-making. He added that AI could make skilled labor more productive while reducing demand for some traditional white-collar tasks.
Despite the uncertainty, Wolfers urged lawmakers not to expect immediate evidence of AI-driven growth in economic data.
“Most people’s work today looks a lot like it did two years ago,” he said.
Wolfers concluded by emphasizing that AI’s ultimate economic impact will unfold over years, not months, and that Michigan’s policy decisions could play a key role in determining whether the technology widens inequality or broadly boosts productivity and growth.
“No one really knows how big this will be,” he said. “But it’s clear this is a technology that matters.”
