Artificial intelligence could automate more than half of U.S. work hours by 2035 without producing an equivalent decline in employment, according to the McKinsey Global Institute. Technological and economic shifts could reduce labor demand by the equivalent of 36 million jobs, while economic expansion—including growth in AI-related industries and services—could generate demand for approximately 41 million. Total employment could therefore rise even as the composition of work changes substantially.
Office and administrative support occupations are particularly exposed, with roughly 80% of current work hours potentially subject to automation. That estimate concerns working time, not the proportion of jobs expected to vanish. Around 11 million Americans may need to leave declining occupations for different ones, while others adapt within their existing fields. In management and law, where judgment, coordination, and accountability are important, AI is more likely to assist workers than replace them outright. Responsibilities such as reviewing AI-generated work and overseeing responsible use may require additional training.
Credentials could complicate these transitions: about 85% of growing employment requires one, with legal requirements accounting for 38% and employer preferences for another 47%. McKinsey suggests reassessing essential qualifications and placing greater emphasis on demonstrated skills and experience. Rather than measuring only AI-related savings, companies should track occupational transitions, internal hiring, and worker outcomes. Finance leaders should also examine whether freed-up time improves customer service, supports innovation and business growth, or enables more complex work, rather than merely shortening existing tasks.
