Artificial intelligence is poised to have a greater impact on Minnesota’s workforce than most other states’, according to a report released last month. But it’s not too late for workers, unions and state lawmakers to erect guardrails that protect jobs and promote fairness and accountability.
Researchers at the Minnesota-based North Star Policy Action think tank found that the state ranks 10th nationally and second in the Midwest for workforce vulnerability to AI, with 500,000 Minnesotans – or 17% of the workforce – at “high risk” of having their jobs altered by the emerging technology.
Those impacts could be positive or negative for workers, but the implementation of AI is more likely to be good for workers, according to the report, if they have a voice in the process. And that means collective bargaining.
Aaron Rosenthal, North Star’s research director and author of the report, said he hopes its publication serves as a “wake-up moment for both unions and legislators.”
“AI is advancing rapidly in both its technological capacity and its workplace adoption,” Rosenthal said. “It’s time that workers are given protections to ensure that its implementation helps but doesn’t harm them, and that unions are aware of the need to bargain over AI in any negotiation.”
AI implementation can be a good thing for workers, employers and the economy, “but it has to be done in concert with worker welfare,” Rosenthal stressed.
An Al revolution?
AI systems are designed to perform complex tasks normally done by humans, like reasoning, decision making and creating. Generative AI systems like Chat GPT create content – text, images, music and more – from a human prompt, while machine learning AI systems analyze data, images or other systems to perform a task.
Unlike computerization and automation, which have impacted workers in blue-collar jobs like manufacturing, AI is more likely to impact workers with higher levels of education, according to the North Star report. AI is already impacting fields like medical research and health care, law and engineering.
The report points to Minnesota’s strong agricultural, software development and construction industries as areas AI is likely to impact moving forward, making the state more vulnerable than most others. But Rosenthal emphasized that AI is “probably going to touch every industry in the state, and it’s that breadth that’s going to require a similarly broad effort by legislators to respond.”
The stakes are high. AI’s potential impact “puts it into a class with only the most significant economic shifts, such as the Industrial Revolution,” during which labor lost a significant share of its income, according to the report.
Without protections for workers, the report warns, AI could “accelerate” income inequality and consolidate more wealth and power among the elites. One AI executive, Sam Altman, already has predicted the technology will create the world’s first trillionaire.
Legislative responses
What should lawmakers be doing in response? The report offers four policy recommendations, and they begin with strengthening workers’ bargaining rights.
Union density is low in many of the industries AI is poised to impact, but collective bargaining is an effective way for workers to “use their expertise to guide an effective employment of AI in their workplace,” Rosenthal said.
Other policy recommendations include worker training and transition programs, regulations governing AI’s use in employment decisions, and a long-term framework – or task force – for keeping tabs on the technology and its use.
State Sen. Erin Maye Quade (D-Apple Valley) co-sponsored a law passed last session to protect Minnesotans from AI-generated deep fakes and non-consensual sexual imagery. This session she hopes lawmakers will prohibit the use of AI in hiring decisions, as a first step toward regulating the technology’s impact on workers.
While the technology may be good at identifying the typical worker an employer has hired in the past, AI won’t be able to account for the racial or other biases that might have factored into those decisions.
“AI knows what, but it doesn’t know why,” Maye Quade said. “AI is built by humans, so it is rife with the same biases that humans carry. The thing about humans is we can be cognizant of our own biases and know that we have them and try to ameliorate that. AI is not going to do that.”
Maye Quade said she is hopeful that the hiring measure will gain bipartisan support in the Legislature. Other AI-related issues lawmakers could consider, she said, include the environmental impact of data centers built to support AI systems across the state and the technology’s use by state agencies.
“There’s only so much a state legislature can do,” Maye Quade acknowledged. “Even between last session and this session, we have seen AI be integrated almost against our will and in more places in our life. We just keep finding out more places.”
Space for unions to act
Some unions, meanwhile, already have taken steps to protect members against negative impacts of AI. In contract negotiations with entertainment studios in 2023, members of the Writers Guild bargained for language that allows them to use AI as a tool for script writing but protects against the use of fully AI-generated scripts.
And last month, National Nurses United – the umbrella organization that includes the Minnesota Nurses Association – put AI concerns at the forefront of its contract campaign covering more than 100,000 California nurses. The union conducted a survey last year that “found nurses repeatedly reported that their own assessments of patients did not match assessments from AI technology deployed in their facilities.”
North Star’s report suggests Minnesota unions should be taking similar action, whether it’s organizing new workers susceptible to AI’s impacts or bargaining to protect their current members.
“There’s an opportunity here for unions across the state to reach out to individuals in industries that are experiencing newly found precarity due to AI and build cross-class, cross-industry solidarity,” Rosenthal said.
“I think there’s a way of looking at this as Luddite panic about technology, but we’re not saying that AI innovation is a bad thing,” he added. “Far from it. It could complement existing work, making jobs more fulfilling by automating mundane tasks. But it could also produce income inequality by cutting wages, leading to job replacement.
“The difference comes down to legislative responses and, most importantly, the ability for workers to organize and bargain around the shape that AI implementation takes.”
Learn more
Read the full report, “Progress and Protection: Aligning AI Innovation with Worker Welfare,” online at northstarpolicy.org.
– Michael Moore, UA editor