AI Jobs Index: 784 Roles Ranked by Risk

A new study from Tufts University’s Fletcher School has revealed which U.S. occupations face the highest risk of job loss due to artificial intelligence (AI). The American AI Jobs Risk Index, compiled by the Digital Planet team, examines 784 occupations, 530 metro areas, 50 states, and 20 industry sectors, ranking them by vulnerability to AI-driven displacement. The research suggests that roles that benefit most from AI productivity gains are often the same ones at greatest risk.

The projections are based on different AI adoption scenarios, rather than current layoffs or employment changes. In a median scenario, roughly 9.3 million U.S. jobs are at risk, with a possible range of 2.7 million to 19.5 million depending on how quickly AI is adopted. These figures provide a predictive view rather than a definitive forecast, helping to highlight sectors and roles that could face significant changes.

At the top of the risk list are writers and authors, with a 57% likelihood of being affected. Computer programmers and web and digital interface designers follow closely, each with a projected 55% risk. Editors face a 54% risk, while web developers are at 46%. Other occupations in marketing, public relations, and journalism also rank highly, with projected job loss rates ranging from 35% to 37%.

Previous studies, such as Stanford’s “Canaries in the Coal Mine” and the Anthropic Economic Index, mainly examined how accessible jobs were to AI. The Tufts index goes further, estimating how likely AI exposure is to translate into actual job displacement. Researchers highlight what they call the “augmentation-displacement link,” showing that roles enhanced by AI productivity are often at higher risk of reduction.

The logic is simple: as AI increases worker efficiency, companies may be able to maintain output with fewer employees. Entry-level and lower-seniority positions tend to feel this effect first, as firms reduce hiring rather than lay off existing staff. Fields like writing, programming, web design, technical writing, and data analysis are particularly susceptible, as tasks in these roles are structured and language-intensive—ideal for large language models to perform autonomously.

Industry-level analysis shows an average vulnerability of 6% across all sectors. However, certain industries are more exposed: Information (18%), Finance and Insurance (16%), and Professional, Scientific, and Technical Services (16%). High-paying roles with large workforces, such as software developers, management analysts, and market research analysts, could experience the largest cumulative income losses, projected at $757 billion annually.

The current analysis does not account for potential AI-driven job creation, which will be included in future updates. It also does not consider mitigating factors like regulatory restrictions, union influence, or occupational licensing. The researchers emphasise that the results are scenario-based and should be interpreted as predictive, rather than absolute, figures.

These findings challenge the assumption that AI use inherently protects jobs. While AI can boost productivity, it can also accelerate workforce reduction if employers do not invest in retraining staff. Dr Craig Froehle from the University of Cincinnati has previously warned that failing to retrain employees could double turnover costs—a concern now quantified by this Tufts study.

The American AI Jobs Risk Index will be updated regularly as AI technology and labour market conditions evolve. Future versions are expected to include both job creation and job loss estimates, providing a more complete picture of AI’s overall impact on employment. The methodology and data are publicly available on the Digital Planet website, offering a valuable resource for businesses, policymakers, and researchers seeking to understand how AI may reshape the workforce.