Firms and Artificial Intelligence: A Regional Update

Regional Matters

February 5, 2026

Similar to past technological developments, the productivity implications, labor market implications, and thus economic implications of Artificial Intelligence (AI) will evolve over time. A lot depends on who is using AI tools, when they are using them, and how they are using them.

In the Richmond Fed's December business surveys - which were fielded between Dec. 1 and Dec. 17 - we asked firms if they have adopted AI and if so, how they were using it. Businesses reported that they were increasingly providing employees with access to AI tools to complete tasks but were less likely to have incorporated AI into their operations. In addition, using data from our June 2024 surveys, we learned that on average, firms were accurate in their assessment of when and how they would start incorporating AI technology into their businesses.

Firms Expectations for AI Use by 2026 Was Generally Accurate

To understand the productivity and labor market implications of the technology over time, we asked firms about the tasks and operations where they were using AI tools. The structure of the questions provides insight into the sort of task-based approach that researchers such as Acemoglu (2025) and Liu et al. (2025) have used to understand implications of technological change over time. In December, 70 percent of Fifth District respondent firms reported that they provided employees with AI tools in their jobs - either public tools or tools provided by their companies. Forty percent of firms provided public tools only, while 13 percent did not allow public tools but provided their own. Meanwhile, fewer firms, but still more than half (56 percent) reported using AI in their operations. For example, firms were much more likely to say that their employees were using AI tools for tasks such as drafting text or generating graphs than for operations such as automating repetitive tasks or managing finances.

In our June 2024 surveys, we asked firms if they were using technology to automate tasks previously done by employees, and if the implemented technology was AI. Unlike our December 2025 questions, the June 2024 survey provided an explicit definition for AI: "computer systems capable of performing tasks that historically only a human could do, such as reasoning, summarizing, making decisions, or solving problems."

Although direct comparisons between expected and actual AI adoption are challenging due to varying questions and changing sample composition, the aggregate data show that firms' 2024 projections for AI adoption by 2026 generally aligned with the actual adoption levels reported at the end of 2025. When asked in June 2024, only 16 percent of respondents had automated tasks with AI at any point in the last two years. However, 45 percent of respondents expected to automate tasks with AI by 2026.

AI Currently Used to Improve Efficiency, Not Replace Labor

Innovations such as electricity, cars, computers, and the internet have fundamentally shaped the skills required by employers. Technology can replace jobs, but it often augments labor or changes the skills required in that job. In our Dec. 2025 business surveys, firms do not think of AI primarily as a way to reduce the need for workers. In response to a question about the importance of factors in adopting AI, efficiency and productivity come up much more often than a reduction of labor costs. When we combine AI investment questions with employment questions, we don't find that firms using AI are more likely to report a decrease in employment in the last month or expect a decrease in employment in the next six months compared to those who are not using AI. Of course, as we report through our monthly employment index, most firms (especially non-manufacturing firms) see and expect employment to be relatively steady.

Firms Are Using AI for Tasks, Not Operations

About half of the respondents who provide AI tools reported using it for tasks like analyzing quantitative and qualitative data, drafting or editing documents, generating graphs or images, or summarizing meeting notes. Even with respect to tasks, fewer reported using it for accounting/finance tasks or customer relationship management. For operations, although about 40 percent of respondents reported using AI to automate repetitive tasks, the operational changes that could engender lasting productivity gains - such as improving worker safety, tracking productivity, optimizing inventory, or assessing and tracking risk - seemed to be less prevalent among respondent firms.

Conclusion

In order for AI technology to enhance productivity in the U.S. economy, the tools have to be adopted by businesses across multiple sectors and incorporated into both tasks and processes. This will take time, but as new and complementary solutions emerge, firms will learn to adapt their operations to take advantage of those solutions. Thus far, although many Fifth District firms have started providing AI tools to their employees for specific tasks, the technology is not yet infused into their business operations. However, responses to our surveys indicate that expectations in 2024 for adoption by 2026 have been exceeded and firms expect continued expansion of the use of AI technology in their businesses and in their industries.

Views expressed are those of the author(s) and do not necessarily reflect those of the Federal Reserve Bank of Richmond or the Federal Reserve System.