New Data Reveals the Tasks Generative AI Is Actually Performing in U.S. Workplaces

As generative AI continues to dominate headlines with promises of transformative change, a new research column from the Washington Center for Equitable Growth offers a rare, data-driven glimpse into reality on the ground. Funded by Equitable Growth, the study leverages a nationally representative survey to measure exactly which workers are adopting generative AI tools — and, crucially, for what tasks.

The analysis moves beyond speculation and broad forecasts to pinpoint the actual work that generative AI is doing in today’s economy. By asking workers directly about their use of tools like ChatGPT, Copilot, and similar platforms, the research sheds light on the types of job functions that are being automated or augmented, as well as the demographic and occupational patterns behind adoption.

Measuring real adoption, not just potential

Much of the early discussion around generative AI focused on which jobs could be affected. The new study flips the script, using representative data to reveal which employees are already using these technologies. According to a column outlining the findings, the research fills a critical gap: official labor statistics have yet to systematically capture generative AI use at the task level.

The survey methodology allows researchers to connect AI usage to specific occupations, industries, and worker characteristics. This granular view is essential for understanding whether generative AI is widening or narrowing the digital divide, and which segments of the workforce may be left behind.

What tasks are workers using generative AI for?

The broad contours point toward a familiar pattern: generative AI is making inroads especially in white-collar, writing-intensive, and analytical roles. Preliminary indications from the research suggest that tasks involving drafting emails, summarizing documents, coding assistance, and data analysis are among the most common use cases.

• Content creation and writing: Workers report using generative AI to draft marketing copy, reports, and internal communications.
• Programming and technical support: Developers and IT staff are employing AI to generate code snippets, debug, and explain technical concepts.
• Administrative and research support: The technology is being used to summarize articles, generate meeting notes, and streamline routine office tasks.

At the same time, the data underscore that generative AI is not yet a universal workplace tool. Adoption rates vary sharply, often clustering among higher-income, college-educated professionals and in specific sectors like technology, finance, and media.

Implications for equity and economic mobility

The Equitable Growth funding context places the research squarely within a debate about inequality and labor-market outcomes. If generative AI boosts productivity primarily for already-advantaged workers, it risks deepening existing disparities. Conversely, if it can be steered toward supporting lower-wage or less-educated workers, it could become a tool for upward mobility.

The column argues that policymakers must pay attention to these adoption patterns if they want to design interventions that promote equitable access to AI’s benefits. That includes considering how generative AI might reshape job design — breaking occupations into component tasks, some of which can be offloaded to AI while others remain distinctively human. For employers, the findings suggest a need to evaluate whether AI tools are being deployed in ways that enhance workers’ skills and job quality, rather than simply replacing labor.

A data foundation for the AI policy conversation

The Equitable Growth-funded study stands as one of the first nationally representative snapshots of generative AI in the American workplace. Its strength lies not in forecasting but in documenting the present — a crucial baseline for tracking future changes. As more workers gain access to these tools, the research can illuminate how adoption evolves and whether it delivers on the productivity gains that many economists anticipate.

With official statistical agencies still working to incorporate AI questions into standard surveys, independent research like this will be essential for understanding the real-world trajectory of generative AI. The findings remind us that while the technology may feel ubiquitous in certain online circles, its footprint across the broader economy is still uneven, and the work it actually does is far from uniform.

For workers, business leaders, and policymakers alike, the message is clear: the generative AI revolution is not one single story, but a patchwork of adoption, shaped by occupation, education, and access. Understanding that patchwork is the first step toward making sure its benefits are broadly shared.