Over the past few years, artificial intelligence (AI) has evolved from a possibility into reality in workplaces and everyday life. Its expanding use is introducing new ways of working, even as its longer-term effects on workers, employers, and communities remain uncertain.
The Federal Reserve’s Community Development function has documented this change and contributed to a growing body of research on the potential economic impacts of AI. This work builds on the Federal Reserve’s longstanding engagement with workforce development and its interest in understanding the forces shaping employment and economic opportunity across communities, including among low- and moderate-income individuals and households.
Many of you might find AI exciting, full of promise and possibility. For others, it may feel more worrisome, a potential disrupter to lives, livelihoods, and what it means to be human.
– Mary Daly, president, San Francisco Fed
February 2026
As often happens during the arrival of new technology, AI’s emergence and expansion have sparked mixed feelings and questions about the future of work. Speaking in San Jose, California on AI’s possibilities, productivity implications, and impact for monetary policy, San Francisco Fed President Mary Daly captured both the enthusiasm and anxiety surrounding AI. “Many of you might find AI exciting, full of promise and possibility,” she said. “For others, it may feel more worrisome, a potential disrupter to our lives, livelihoods, and what it means to be human.”
The questions Federal Reserve researchers and practitioners are examining about AI and work reflect this tension. Which workers and occupations are most likely to experience change? What skills will workers need? How are employers and workforce institutions responding? And which communities are positioned to benefit?
This article shares what community development teams across the Federal Reserve System have learned so far.
Impact: How is AI changing work?
Workers, employers, educators, and communities all want to understand what AI means for jobs, skills, and economic opportunity. While many long-term effects remain uncertain, Federal Reserve research offers early insights into how AI is reshaping the labor market along several dimensions.
Augmentation versus replacement
One of the most fundamental questions accompanying the arrival of new technologies, including artificial intelligence, is whether machines could ever fully replace humans and the work they do. The third issue of the Boston Fed’s Invested magazine tackled this question head on. The chapter on human-machine collaboration highlighted the need for humans to intentionally design both technology and policies that are complementary and present the least disruption to workers and businesses. An examination of how technology and humans might compete for work featured perspectives from cocktail servers, journalists, and a variety of other workers about AI’s potential impacts on their jobs.
We are in a time of rapid technological advancement, and we must adapt to it… I’m partly worried about it too, but we also have to move forward with technology, because sometimes it helps a lot.
– Focus group participant, Federal Reserve Worker Perspectives Project
People expressed uncertainty around new technology’s impact on their jobs in the Federal Reserve’s Worker Perspectives report. The Worker Perspectives project offers a unique view into how job seekers and workers in low- and middle-wage occupations experience the labor market. In focus groups, participants’ views ranged from concern about the pace of technological change and its impact on their jobs to optimism around its capacity to increase productivity and help in the workplace. Workers’ experiences point to the immediate, practical question of better understanding which workers and occupations are most likely to experience change in the first place, and what that change could look like for them.
Job exposure
Rather than predicting whether AI will ultimately create or eliminate roles, job exposure studies estimate how much AI could affect the underlying tasks within different occupations. While there appear to be potential implications for individuals across a range of income and education levels, many studies rank jobs that require a college degree and pay relatively well (for example, computer programmer) at the top of exposure rankings. However, community development researchers highlight how low- and middle-wage jobs may be impacted as well.
Dallas Fed and Philadelphia Fed job exposure studies examine AI’s potential occupational impacts at a regional level. Both find that administrative and technical roles are more likely to be impacted by AI, with the Dallas Fed study also highlighting that positions in the skilled trades will be relatively less exposed. On-the-job AI exposure research from the San Francisco Fed pays particular attention to lower-income workers, finding that members of lower-income households account for 20 percent of US workers highly exposed to AI. That figure can be further broken down by state to better understand local exposure patterns. The research showed that these individuals are more likely to hold office and administrative support roles or work in service-oriented industries like healthcare and social assistance.
Each of these studies emphasizes that high exposure to AI does not necessarily mean a worker is more likely to be replaced. It could also mean that incorporating AI could lead to increased productivity and even wage growth in certain occupations. The same exposure can lead to different outcomes across the labor market, a Minneapolis Fed study found. Impacts depend on the tasks involved in a given occupation, how AI continues to evolve, and how the technology gets integrated into jobs as employers seek increased productivity and output. Continuously updated job exposure studies help stakeholders better understand how AI might reshape different occupations.
Employers’ demand for AI skills
Employers’ hiring decisions, changing skill requirements, technology adoption, and investments in workforce training reveal both where AI is being used today and how its role in the workplace is evolving. To better understand where AI-induced job changes may already be occurring, researchers examine job postings, explore public and proprietary labor market information, and learn from direct engagement with business owners and workers.
As businesses experiment with and integrate AI into the workplace, job postings provide an early indication of where and how employers, human resources departments, and job recruiters indicate demand for AI-related skills, such as machine learning and natural language processing. Drawing on job posting data, Atlanta Fed research found that AI demand varies considerably by education, occupation, and geography. While demand for AI-related skills has steadily increased since 2010, just three percent of online job postings requested proficiency in at least one of these skills. This growing but modest demand also is not evenly distributed. Focusing on six states in the Southeast, Atlanta Fed researchers found AI-related hiring has increased throughout the region, but demand remains concentrated in larger metropolitan labor markets. Further job posting research found that demand for AI skills is greatest for positions that require a bachelor’s degree or more but is growing for jobs requiring associate degrees.
Adoption and implementation
Additional Federal Reserve studies provide another view into employer demand by tracking businesses’ AI adoption and implementation.
Richmond Fed research found that businesses adopt AI at different rates and for different purposes, reflecting variations in industry, organizational readiness, and resources. A St. Louis Fed study on the use of AI in the healthcare workplace found hospitals have begun to integrate AI into their work. AI use is more common in urban hospitals, though, and for particular uses like optimizing administrative and clinical work as well as automating tasks.
Similarly, the Federal Reserve’s 2026 Report on Employer Firms, with results from the 2025 Small Business Credit Survey, showed almost half of firms or their employees use AI, with another 15 percent of firms planning to begin using AI in the next 12 months. Notably, a third of firms had no plans to use AI. A San Francisco Fed report on the 2024 Small Business Credit Survey details the wide range of qualitative responses to questions on AI, finding “sentiment toward AI ranged from enthusiasm to indifference to skepticism.”
Adaptation and response: How are workers preparing for change?
Research shows AI has started to change both how work is performed and which skills employers seek. How can workforce and education systems prepare workers to adapt?
Evidence so far points to an emerging shift in workforce development. Workforce organizations and educators now are doing more than just training a relatively small segment of the workforce to specialize in advanced AI. Increasingly, programs help workers across many occupations build practical AI literacy alongside the foundational, technical, and durable human skills needed to succeed in a changing labor market.
The benefit of AI to society as a whole will depend on the adaptability of workers’ skills, how well they are retrained or redeployed, and how policymakers choose to support the groups that are hardest hit by these changes.
– Governor Lisa D. Cook, Member of the Board of Governors of the Federal Reserve System
September 2023
The importance of this adaptation extends beyond individual workers, with broader implications for how the benefits and disruptions of AI are experienced across society. In a 2023 speech on generative AI, productivity, and the labor market, Federal Reserve Governor Lisa D. Cook highlighted the importance of worker adaptability and the systems that support workers through technological change. “The benefit of AI to society as a whole,” she said, “will depend on the adaptability of workers’ skills, how well they are retrained or redeployed, and how policymakers choose to support the groups that are hardest hit by these changes.”
Education, workforce, and training systems
Educational and workforce training institutions face the challenge of helping workers adapt to a rapidly evolving technology while also learning how to embed AI technologies into their own operations. These institutions must balance meeting employer skill demands with making decisions about curricula, faculty development, funding and partnerships, and access.
As noted above, employers are increasingly requesting that workers start their jobs competent in relevant AI skills. How institutions translate that demand into education and training is still taking shape. These themes closely mirror the Atlanta Fed’s conversations with community colleges across the Southeast. Ongoing dialogue reveals that many colleges are experimenting with how to incorporate AI into curricula, teaching, operations, and workforce programs. Approaches vary, but a common theme is that AI may become a foundational workplace competency that is integrated across all disciplines. Educators are also taking different approaches to equipping students with emerging skills and certifications, offering everything from AI competency modules and industry, academic, and micro-credentials to full AI-related degrees.
Capacity and implementation challenges
Yet preparing workers requires more than developing new curricula. Recent San Francisco Fed community engagement highlighted a landscape of varying adoption, with public sector organization professionals sharing insights from the growing field of AI training and skill development. They observed that AI adoption has been uneven across businesses, nonprofits, and workforce organizations, with many institutions balancing growing interest in AI with limited capacity and uncertainty about implementation.
Organization staff commonly described using AI to improve productivity, backfill hard-to-recruit positions, or improve efficiency amid resource and budgetary constraints. At the same time, stakeholders pointed to persistent adoption barriers, including limited access to technical assistance and concerns about privacy and responsible use.
They also expressed concern that existing digital divides could widen as AI proliferates. These constraints may be particularly consequential for institutions and communities with fewer resources for AI experimentation and adoption.
These challenges underscore the reality that preparing workers for an AI-enabled economy is also a question of institutional capacity. Workforce development and educational organizations will need the ability to continually update programs, respond to changing employer needs, and build partnerships that can support sustainable technology and AI-enabled talent pipelines.
Looking ahead: Questions for the future of work
Preparing workers with the skills to succeed in an AI-enabled economy is only part of the challenge. The extent to which that preparation translates into economic opportunity also depends on the jobs available where workers live, their access to technology and training, and the capacity of communities to connect workers with emerging opportunities.
Communities also enter this period of technological change with different resources and capacity to respond. Larger metro areas tend to have more AI demand and a broader employer base; many smaller and rural markets have less. Broadband access and digital literacy remain important foundations, but capacity extends beyond access to technology. Some places benefit from dense networks of employers, as previously mentioned, but also from educational institutions, economic development groups, and other intermediaries and system stakeholders that can identify and coordinate the right set of responses to connect workers with employer needs. As a result, existing differences in local capacity could influence how quickly communities are able to translate technological change into new pathways for workers.
Federal Reserve community development work has provided insight into how AI is affecting work and how employers, workforce organizations, and communities are responding so far. But the pace and distribution of those changes remain uncertain, raising a new set of questions for the next phase of research and engagement around AI and workforce development.
How quickly will AI reshape work?
A gradual transition would allow employers, educators, and workforce systems more time to adapt, while a more rapid transformation could challenge institutions’ ability to prepare workers and support communities through change.
How should workforce systems prepare workers?
While researchers continue to study AI’s effects on occupations, early evidence suggests the impacts will not be uniform across regions, types of workers, and industries. What combination of AI literacy, occupational expertise, durable human skills, and lifelong learning will best prepare workers for long-term success? And how can employers, educators, and workforce systems continue to build accessible pathways to quality jobs and economic mobility for all workers?
How might we measure AI’s impact on workers and communities?
Measures of employer adoption, exposure scores, and job postings help track where AI is emerging, yet they reveal less about whether workers and communities are benefiting. Beyond counting job postings, tech adoption, or AI credentials earned, how should we define success? What are the best ways to measure successful adaptation, including skill development, career progression, job quality, and long-term economic mobility?
It remains to be seen how AI is actually going to be used in the future. And of course, there’s speculation that it’s going to replace the manpower and replace human beings, which there’s still the human touch, which is going to be necessary in general.
– Focus group participant, Federal Reserve Worker Perspectives Project
How can AI expand opportunity without widening existing disparities?
Ensuring that AI benefits businesses, workers, and communities across the country will require more than expanding access to technology. Questions remain about how digital literacy, broadband availability, institutional capacity, and differences between urban and rural economies may hamper any benefits from AI. Ongoing work can help identify what investments in partnerships, technical assistance, and workforce systems will be needed to help everyone participate in an AI-enabled economy.
The answers to these questions will become clearer as AI adoption continues and the technology itself evolves. What is already apparent is that AI’s impact on work will not be determined by technology alone. They will also reflect the choices of innovators, policymakers, and each member of the workforce development ecosystem: employers, workers, educators, workforce organizations, and communities.
Continued research and engagement can help untangle the implications and opportunities of AI, helping stakeholders steer toward broad-based economic growth and greater opportunity for workers throughout the labor market.
The views expressed herein are solely those of the authors and should not be reported as representing the views of the Federal Reserve Banks of Atlanta or Dallas, the Board of Governors, or the Federal Reserve System.
