2025 reminder for 2026: Microsoft study on who must adapt fast to AI

Communicate Staff

January 2, 2026

As 2026 begins, a Microsoft Research study published last year is being revisited by policymakers, employers, and workers alike, offering a data-backed snapshot of which jobs are most likely to be affected by generative artificial intelligence—and why adaptation, rather than alarm, should be the focus of the professionals in these jobs.

The study, Working with AI: Measuring the Occupational Implications of Generative AI, was released in mid-2025 by a team of Microsoft researchers and quickly became a reference point in debates about AI and employment. Its findings are gaining renewed attention as generative AI tools, including Microsoft Copilot and other large language models, have become more embedded in everyday workplace software since July 2025, when the study was released.

Rather than predicting job losses, the researchers focused on how AI is already being used. The analysis examined more than 200,000 anonymized, real-world interactions with Bing Copilot, studying the tasks users asked AI systems to perform. These activities were then mapped against job tasks listed in the U.S. Department of Labor’s O*NET occupational database.

From this, the researchers developed what they called an “AI applicability score,” measuring how often and how effectively generative AI could assist with key tasks within a given profession. High scores indicate roles where AI overlaps significantly with daily work activities, such as writing, summarizing information, translating languages, analyzing data, and responding to customer inquiries.

The authors stressed that the results should not be interpreted as forecasts of widespread job displacement. “This is not a prediction of automation or replacement,” the paper noted, emphasizing that many tasks still require human judgment, accountability, and contextual understanding. Instead, the findings highlight where work is most likely to change.

As generative AI use accelerates, that distinction is becoming clearer. Many of the professions identified as highly exposed—including journalists, writers, customer service representatives, analysts, and software developers—have already seen AI tools integrated into core workflows over the past year.

Labor economists say the study stands out for grounding its conclusions in observed behavior rather than theoretical modeling. “This research is valuable because it looks at what people are actually doing with AI today,” said a technology policy expert familiar with the paper. “It signals where workers may need to upskill or rethink their roles, not where jobs will vanish overnight.”

For workers in high-applicability professions, staying relevant increasingly means learning how to work alongside AI—by refining prompts, verifying outputs, applying domain expertise, and exercising editorial and ethical judgment. Roles are shifting from producing first drafts to supervising, contextualizing, and validating machine-generated content.

The researchers framed generative AI as a productivity-enhancing tool rather than a standalone substitute for human labor, likening it to a “copilot” that reshapes expectations around speed and scale. Employers, they noted, will play a key role in determining whether these tools augment human work or intensify pressures on workers.

As governments and companies continue to debate how to regulate and deploy AI in 2026, the Microsoft study serves as a reminder that the impact of generative AI is already measurable—and that preparedness, not panic, will shape how workers fare in the years ahead.