Generative AI and White-Collar Work

How AI systems that write, code and analyse are changing office jobs, which tasks are most exposed, and what early studies show.

Earlier waves of automation mostly affected routine manual and clerical tasks. Generative AI, which can produce text, images, code and analysis, affects a different set of jobs: many white-collar, knowledge-based roles once thought safe from automation.

Which tasks are exposed

Researchers measure exposure by looking at how many tasks in a job AI could help perform. A 2023 study by OpenAI and University of Pennsylvania researchers estimated that around 80 percent of U.S. workers had at least some tasks exposed to large language models, and that higher-paid jobs tended to be more exposed. Tasks like drafting documents, summarising information, writing code and answering customer questions are highly exposed. Physical and in-person tasks are less so.

Replace or assist?

Exposure does not necessarily mean replacement. AI may:

Automate some tasks entirely.
Augment workers, making them faster or better at their tasks.
Create new tasks and jobs, such as working with and checking AI systems.

The International Labour Organization found in 2023 that augmentation was more likely than full automation for most occupations, with clerical work the most exposed to automation.

Early evidence

Studies have found productivity gains in specific settings. Research by Erik Brynjolfsson, Danielle Li and Lindsey Raymond at a customer support centre found that access to an AI assistant raised the number of issues resolved per hour by about 14 percent on average, with the largest gains for less experienced workers. Experiments with professional writing and coding tasks have also found faster completion times.

Two new analysts

A company hires two junior analysts. With an AI assistant, they draft reports in half the time, and their work quality comes closer to that of senior colleagues. The company may need fewer junior analysts for the same work, or it may take on more projects. Which happens depends on demand for the company's services.

Open questions

Economists debate how quickly AI will spread, how much it will raise productivity across the economy, and how the gains will be shared between workers, companies and consumers. The history of general-purpose technologies suggests effects may take years to appear fully.

Thinking exposure means a job will disappear

A job with many AI-exposed tasks may change a great deal without disappearing. Workers often spend time on tasks AI cannot do, and demand for their services may grow if AI makes them more productive.

Key takeaways

Generative AI affects many white-collar tasks, including writing, coding and analysis.
Studies estimate a large share of workers have some tasks exposed to AI.
AI may automate tasks, augment workers or create new tasks.
Early studies found productivity gains, often largest for less experienced workers.