A new framework from OpenAI suggests the conventional wisdom about AI and jobs is too simple. Rather than asking which jobs AI can do, the framework considers three things: Can AI handle a meaningful portion of the tasks? Does a person still need to be in the loop for oversight or accountability? And if AI lowers costs, will demand increase enough to keep people employed? Applied to 148 million U.S. jobs across 921 occupations, the model shows roughly 18% of jobs are at high automation risk (think data entry and telemarketing), 24% will likely reorganize (like lawyers and accountants keeping oversight roles), 12% could actually grow thanks to AI, and 46% show less immediate change. Why it matters: This paints a far more nuanced picture than the usual “robots are coming for your job” narrative—the real story is about work transforming, not vanishing overnight.