Миннесота адаптирует экономику штаб-квартир к вызовам искусственного интеллекта

Several years ago, I spoke to Jawed Karim, a former St. Paul resident who co-founded YouTube before Google purchased it in 2005. A soft-spoken computer engineer who is part Bengali and German, Karim somehow escaped the celebrity spotlight of the early internet entrepreneurs who made their fortunes during the first decade of the 21st century. Karim was in town to promote his new venture capital fund when he said something about Silicon Valley and the Midwest that has stayed with me ever since. “In Silicon Valley, if people hear that you’re working for Oracle or Cisco, they will feel bad for you and say, ‘Hey, I can help you get a job at a startup,'” Karim told me. “In Minnesota, if people hear you’re working for a startup, they will feel bad for you and say, ‘Hey, I can help you get a job at 3M.'” At the time, I took his observation as a critique of the Midwest’s lack of risk tolerance, one reason Minnesota has struggled to attract venture capital and launch tech startups in significant numbers. But over the years, I began asking a different question: What if Minnesota’s economic model wasn’t broken? What if it was simply different? Rather than producing hundreds of venture-backed startups, Minnesota built one of the nation’s strongest headquarters economies. An unusually high concentration of Fortune 500 companies has generated high-paying jobs, billions in investment and a remarkable tradition of corporate philanthropy. That model has served the state extraordinarily well. However, Minnesota’s greatest economic strength could also become its greatest weakness. The rapid emergence of generative artificial intelligence, including ChatGPT, Gemini and Claude, threatens many of the white-collar professions on which Minnesota’s headquarters economy depends: lawyers, engineers, communications professionals, accountants, market analysts and sales representatives. “AI represents a particular vulnerability to our economy,” said Aaron Rosenthal, research director for North Star Policy Action, a local left-leaning think tank. “We tend to have a disproportionate number of white-collar jobs that’s more susceptible to AI.” Rosenthal, along with researchers at the University of St. Thomas, recently authored a report estimating that one-third of Minnesota’s labor force — roughly 800,000 workers — are employed in occupations with “high levels of exposure” to AI. So far, there is little evidence that AI is replacing workers in significant numbers. But given the speed and scale of the technology, it would be naive to assume the state can indefinitely resist its disruptive pull. “This is the Second Industrial Revolution sped up to the max,” Rosenthal said. That raises a larger question: If AI dramatically increases productivity, who will capture those gains? Will Minnesota’s corporate giants use AI primarily to reduce costs and boost shareholder returns? Or will they use it to launch new products, create new business lines, invest in new ventures and generate the next generation of Minnesota jobs? Headquarters economy That Minnesota has been home to so many Fortune 500 companies remains something of a mystery, even to experts who have studied the issue. “It’s not like we were just gifted all of these headquarters,” Myles Shaver, a professor of corporate strategy and strategic management at the University of Minnesota, said in an interview. “It was a confluence of different things.” In his book Headquarters Economy, Shaver points to the Twin Cities’ location along the Mississippi River and its proximity to Canada, advantages that helped give rise to the state’s mining, agriculture, and milling industries. But even Shaver acknowledges that geography alone cannot explain why Minnesota became home to such a diverse collection of industries, including retail, food, health care and financial services. In addition to publicly traded corporations like Target, 3M, General Mills, U.S. Bancorp and UnitedHealth Group, the state is home to some of the nation’s largest private companies, including Cargill and Thomson Reuters. Non-proft Mayo Clinic, one of the country’s largest hospital and medical centers, is based in Rochester. Locally grown firms that were later acquired, such as Medtronic and Supervalu, still maintain major operations here. That diversity helped Minnesota avoid the fate of manufacturing-dependent Midwestern cities like Detroit, St. Louis and Cleveland. It also made the Twin Cities an attractive destination for generations of white-collar professionals despite its famously harsh winters. One thing, however, is clear, Shaver said. As more headquarters clustered in one region, they created an ecosystem of administrative and managerial talent that could move easily across companies and industries. Over time, Minnesota’s comparative advantage shifted from producing goods to producing organizational capability. A state that began with farmers and millers gradually developed a workforce skilled at planning, coordinating, allocating resources, solving operational problems and executing strategy across increasingly complex organizations. Resilience vs. creativity As a result, Minnesota’s economy has come to reward managerial excellence perhaps more than entrepreneurial risk-taking. Last year, Minnesota startups in their first year of existence created just 3.91 jobs per 1,000 people, according to the Kauffman Foundation, compared with the national median of 4.8 jobs. Only seven states performed worse. Minnesota also ranks a modest 20th in attracting venture capital. Not every state needs to resemble Silicon Valley, of course. Minnesota may not create as many new companies as other regions, but “we build businesses that last,” said Cameron Macht, regional analysis and outreach manager for the Labor Market Information Office at the Minnesota Department of Employment and Economic Development. However, resilience is not the same thing as growth. For decades, Minnesota’s greatest strength has been its ability to execute — to build durable organizations that continually improve and adapt. Over the past decade, Minnesota’s economy has grown about 1.7% annually, compared with 2.4% nationally. Since the pandemic, yearly job growth has averaged just under 1%. The challenge is that AI is becoming increasingly capable of performing many of the very functions that headquarters have historically existed to perform: scheduling, forecasting, writing reports, synthesizing information, coordinating workflows, drafting plans and tracking projects. These are the kinds of execution-oriented, managerial tasks that have long defined Minnesota’s headquarters economy. What AI has yet to consistently master are some of the functions at the other end of management: deciding which opportunities are worth pursuing, inspiring people around a common vision and creating entirely new businesses. In other words, strategy and leadership. Built to last Minnesota companies have displayed plenty of those traits — when circumstances demanded it. New technologies have disrupted industries before, but local companies have consistently endured and even flourished. “Minnesota is the story of companies constantly reinventing themselves,” Shaver said. The state may lack outside entrepreneurs, but there’s “plenty of innovation that happens within companies,” he said. Moreover, Minnesota companies know how to adapt without losing continuity with the very things that made them special and successful in the first place. But such situations have been messy. Take Best Buy. As detailed in my book Rebuilding Empires, the Richfield-based consumer electronics giant in 2011 was in deep trouble due to market share losses to Amazon. Rival Circuit City had filed for bankruptcy, and it looked like Best Buy would soon meet a similar fate. At the time, Best Buy fired its CEO and replaced him with a board director who wanted to quickly close stores and shrink the retailer’s footprint. That prompted founder Richard Schulze to launch a bid to take the company private. Eventually, Best Buy hired Hubert Joly as CEO. Joly realized the company’s problem wasn’t store space but how the retailer used it. So rather than close stores, Joly introduced buy-online, pick-up-in-store service and store-within-a-store concepts from major vendors like Sony and Samsung. He also matched Amazon prices and reduced lower-margin items like DVDs. The internet was supposed to wipe out physical shopping. Instead, retailers like Best Buy and Minneapolis-based Target held their ground and became formidable e-commerce players on their own. Different game Some local experts believe AI will cause upheaval in Minnesota’s headquarters economy but, just like the internet, the state has the necessary leadership and managerial talent to adapt. I’m less optimistic. Yes, technologies have come and gone throughout history, and we’re still here. I’m sure some horse lovers in the late 19th century thought the world was going to end when the first cars started to rumble across American streets. But AI is different for two reasons. The first is that we’re not exactly sure what it is capable of becoming. Programs like Claude and ChatGPT remain something of a black box. It took corporations years to figure out nonphysical abstractions like the internet, and some are still struggling. The second is what truly frightens me: the speed at which AI capabilities are improving. The internet was enormously disruptive, but it was fundamentally a tool operated by humans. Today’s AI systems are increasingly capable of completing complex sequences of work with less human intervention. One nonprofit research organization that has gained increasing attention among AI researchers is Model Evaluation & Threat Research, which seeks to measure what the latest AI systems can actually do. METR evaluates whether AI can successfully complete increasingly long and complex real-world tasks with minimal human intervention. The organization refers to this as an AI system’s “task horizon” — the amount of work it can reliably accomplish on tasks that would take a skilled human a given amount of time. SUBSCRIBE: GET THE MORNING HEADLINES DELIVERED TO YOUR INBOX. METR’s findings suggest that frontier AI systems are improving at a striking pace. According to the organization’s research, the length of tasks that leading AI models can complete has roughly doubled every seven months since 2019. While that doesn’t mean AI is becoming twice as intelligent every seven months, it does suggest that the systems are steadily acquiring the ability to perform longer, more sophisticated sequences of knowledge work — from tasks measured in minutes to those requiring hours of sustained reasoning and coordination. Corporate response A lot of people are naturally worried about AI eliminating jobs, especially politicians. Earlier this year, Gov. Tim Walz proposed a tax on social media companies and using the proceeds to help workers impacted by AI. Another bill proposed requiring companies to provide 90 days’ notice before deploying technology that could displace jobs. During this transitional work period, employees would continue to receive regular pay and would be offered a chance to reskill or upskill. Luke Greiner, regional analyst for Central and Southwest Minnesota at DEED, thinks AI will actually help address a perennial problem in Minnesota’s economy — the lack of workers — by helping the current workforce become more productive. “We don’t have enough people to grow the economy,” Greiner said. In fact, he suggests the primary issue will be how workers individually adapt to AI. Some kind of safety net and training assistance for workers makes sense, but these are purely defensive measures. What about the scenario in which businesses use AI to create new markets, industries and jobs, just as the printing press, steam engine and internet have done? Unfortunately, this is where not having more Jawed Karims and well-funded high-tech startups willing to take big risks really hurts. Minnesota may still produce breakthrough technologies through institutions like the University of Minnesota and Mayo Clinic. But most startups fail, which is why an economy needs a lot of them to increase the chances that some will succeed. No, Minnesota’s future with AI will largely depend on the Fortune 500 companies of the headquarters economy. “Can companies drive more revenue without driving workers out of jobs?” Rosenthal said. “Or do they focus on ‘how do I squeeze more profits from automation?’ We don’t know.” Minnesota companies are certainly trying to figure it out. UnitedHealth plans to invest at least $1.5 billion in technology and AI this year and again in 2027 to “help make health care simpler, more affordable and easier to navigate,” the company said in an email. “We are using AI-enabled modernization to help make health care simpler and more connected, improve experiences for consumers and care providers, increase productivity and reduce administrative burden,” it said. Jeff Shelman, a spokesman for U.S. Bancorp, referred me to the company’s annual report. “We’ve increased efficiency and enhanced client experiences by investing in AI and machine learning,” the report said. “For example, our engineers use AI tools to write and review code faster and to automate processes. Our GenAI-powered code-review assistant reduces review time by 75%, so our engineers can focus on higher-value work.” So there is a lot of discussion about AI improving customer experience and boosting efficiency. But there are not many clues to the larger question Minnesota needs to answer: Will AI become primarily a tool for cutting costs, or will companies use those productivity gains to create something new? Because the future of the headquarters economy may depend on the answer.