Headlines about the impact of AI often swing between utopian predictions of unprecedented prosperity and dire warnings of mass unemployment. But for lower-income Americans, the real question is much more practical: Will AI create new opportunities, or will it widen the economic divides that already exist?
According to economist and MIT researcher Neil Thompson, the featured speaker at a Brookings Institution webinar last week, the answer is likely to be complicated. AI will produce both winners and losers, and the choices policymakers make over the next several years could determine whether lower-income communities benefit from the technology or bear its costs.
One reason for optimism is that AI has the potential to lower the cost of services and open access to tools that were previously available only to people with money or specialized expertise. Thompson argued that AI could make many services dramatically more accessible because "you can do much a more exhaustive search because that exhaustive search has become so much more cost effective."
Imagine a parent trying to find the best school for a child with specific learning needs. In the past, that kind of individualized guidance might have required expensive consultants or extensive research. Similar advances could improve access to healthcare information, financial planning, and other services that often remain out of reach for lower-income households.
AI may also create new paths into occupations that once required years of specialized training. Thompson noted that previous waves of automation often reduced barriers to entry into certain professions. GPS technology, for example, diminished the importance of mastering every street in a city, allowing many more people to become drivers.
The same dynamic could occur in healthcare and other fields. AI tools could enable nurses, technicians, and other workers to perform tasks that once required higher levels of expertise. For people without advanced degrees, that could mean access to better-paying jobs and greater economic mobility.
Moreover, Thompson emphasized that technological change does not simply destroy skills—it creates new ones. "Just because certain existing forms of expertise go away," he said, "that doesn't mean that new forms of expertise won't be created." For lower-income Americans seeking upward mobility, the challenge will be identifying and preparing for these new opportunities.
But these possibilities come with significant risks.
Historically, major technological shifts have often created painful disruptions in labor markets. Certain industries, regions, and workers can be hit hard when technology changes faster than people can adapt. Lower-income Americans are particularly vulnerable because they often have fewer financial resources, less job security, and limited access to retraining opportunities.
Importantly, Thompson argues that AI's effects are unlikely to fall neatly along traditional lines such as white-collar versus blue-collar work. Instead, the impacts will vary occupation by occupation. Some workers will see their productivity and wages rise. Others may find that key aspects of their jobs have been automated, reducing demand for their skills.
The biggest danger may be geographic concentration. Entire communities that depend heavily on a single type of work—such as customer service, data processing, or routine administrative tasks—could face severe disruption if AI rapidly automates those activities. Thompson warned that places heavily dependent on a single industry could experience profound disruption, much like communities hollowed out by previous economic transformations.
There is also a broader concern about inequality itself.
As AI systems become more capable, a growing share of economic value could flow to those who own the technology and the capital behind it. If machines increasingly perform work that humans once did, the returns may accrue disproportionately to investors and technology companies rather than workers.
Thompson worries about a future in which "more and more of the economy" accrues "to the owners of capital because more and more of the work is being done by those systems." He added that this could fundamentally challenge our traditional notions of economic equality because "so much of equality that we get is enforced by the fact that we all have one person's worth of labor."
For lower-income Americans, that raises a troubling possibility. Even if AI makes the economy more productive overall, the benefits may not be shared broadly. The nation could become wealthier while economic inequality deepens.
To address that possibility, Thompson expressed openness to policies such as broad-based ownership models or sovereign wealth funds that would allow ordinary Americans to share in the returns generated by AI-driven growth. Such approaches, he suggested, may prove more durable than simply sending government checks.
The good news is that the future is not predetermined.
One of Thompson's most important findings is that AI's impact is likely to resemble a "rising tide" rather than a sudden tidal wave. Instead of entire occupations disappearing overnight, technological change may arrive incrementally, giving workers, businesses, and policymakers time to prepare. "We can look ahead," he said. "Let's plan for a transition of some of those workers to make sure that they're going to be in a good place."