The rise of artificial intelligence could make Boston rush hours a lot less hectic over the next five years — by throwing thousands of people out of work.
A startling new study from Tufts University says that more than 207,000 Boston-area workers and 260,000 statewide are likely to lose their jobs to AI systems over the next five years, resulting in $25.6 billion in lost wages, a major blow to the local economy.
The American AI Jobs Risk Index finds that Massachusetts is the US state with the highest risk of AI job losses. Only workers in the District of Columbia are in greater peril. Greater Boston ranks sixth among US metropolitan areas for expected job losses.
The news may be shocking but it’s hardly a surprise. “What are the kinds of jobs that are most vulnerable to AI?” said the report’s lead researcher Bhaskar Chakravorti, dean of global business at Tufts’ Fletcher School of Law and Diplomacy. “It’s jobs that involve coding, that involve writing, that involve content creation, that involved data analysis, that involve market research, that involve knowledge creation. Guess what? Boston is dense with these kinds of jobs.”
But Boston won’t suffer alone. The Tufts report predicts 9.3 million lost jobs nationwide over the next five years. If that happens, the wages lost to US workers could equal the gross domestic product of Belgium.
All these numbers assume that businesses and governments adopt AI at a moderate pace. Chakravorti said that companies may adopt AI more slowly, leading to much lower job losses. Or they might adopt AI as quickly as possible. In that case, the United States might lose 19.5 million jobs by 2031, with Massachusetts losing 535,000. And that would blow a hole in total US wages as big as South Korea’s entire economy, according to Chakravorti’s report.
There’s a glimmer of hope. Chakravorti’s report doesn’t take into account any new jobs created by the adoption of AI. “I could see a lot of startups coming on board,“ he said. But Chakravorti added: ”They won’t need to hire tons of people. And even with a handful of staff, they can get to a billion dollars in revenue.” So new AI-centric companies aren’t likely to generate as many jobs as they eliminate.
Besides, Chakravorti said, any such bounceback won’t happen quickly enough to make up for all the job losses.
Who gets laid off first? Writers and editors are at the greatest risk, the report says — 57 percent of such jobs are destined for the chopping block. Maybe they should have learned to code.
But the coders come in at second place, with 55 percent likely to be pink-slipped. And the hits keep coming. Mathematicians, website developers, database architects, even atmospheric and space scientists are all in big trouble.
At least journalists catch a break. They’re broken out in a separate category from other writers and given a decent shot at survival. Only 35 percent of these jobs will be AI’d to death.
Whose jobs are secure? “The three safest jobs are if you are a human dishwasher, if you’re a floor finisher, or if you are a technician in a surgical room,” said Chakravorti. He agreed that AI-powered robots can polish floors, but notes that such machines are still quite expensive compared to a human earning minimum wage. In short, there are two kinds of safe jobs — those that AI can’t do at all, and those it can do, but at too high a price.
That’s good news for those who work with their hands. Welders, stonemasons, meat packers, railroad workers, and crematory operators are among the safest jobs around. Even fast food workers are considered at little risk of AI replacement. But if you work mainly with your brain, look out.
How long before the tsunami arrives? A team at the Massachusetts Institute of Technology’s Sloan School of Management is trying to figure it out. They’ve developed a new way of analyzing work, to help predict which jobs are most vulnerable to AI disruption.
Their research, available online but not yet published in a peer-reviewed journal, may someday guide college students into safer career choices, but it might also help employers identify new opportunities to replace workers with AI.
According to Thomas Malone, the MIT Sloan professor leading the research team, answering such questions requires a clearer understanding of work.
“Instead of focusing primarily on jobs,” said Malone, “we focus on the detailed activities that people and machines do at work.” This means creating a catalog of the specific tasks involved in any given job.
Some tasks require acting on information — writing memos and reports, for instance. Others require acting on physical objects, such as loading boxes into a truck, and still other jobs require interactions with others, such as holding a Zoom meeting.
From these three generic activities, Malone’s team teased out thousands of very specialized tasks. The result is a database that should make it easier to predict how AI will work its way through the economy, and its impact on the labor force.
Next, they worked with There’s An AI For That, a popular website with a constantly updated index of available AI apps for consumers and businesses. The team collected information on more than 13,000 such apps, and ran the data through GPT 5.1 to figure out exactly which tasks can be performed by each app. Human researchers double-checked the results.
In addition, the team collected data from the International Federation of Robotics on nearly 21 million robots worldwide, used for everything from cleaning floors to welding metal.
By matching all this information against the database of 19,000 worker tasks, Malone’s team can get a handle on how deeply AI has penetrated the US workplace, and which jobs will be the next to fall.
So far, only 25 percent of AI dollars have gone into systems for controlling robots, mostly for household floor-cleaning machines such as the iRobot Roomba. Malone blames the physical limitations of today’s robots, which still can’t match the dexterity of human hands.
AI mainly threatens workers who manage information. But not all of them. Malone noted that some industries demand human empathy, a sense of ethics, and a knack for teamwork. That’s why he thinks health care jobs are relatively safe.
The Tufts and MIT reports join a tidal wave of similar studies from Stanford University, Yale University, Goldman Sachs and the World Economic Forum. At least AI is creating a lot of jobs for economists.
Hiawatha Bray can be reached at hiawatha.bray@globe.com. Follow him @GlobeTechLab.