Overview
- Drawing on nine months of Copilot interactions, Microsoft decomposed jobs into tasks to measure what share can be automated in real-world use.
- The report names ten occupations at high risk over the coming decade: interpreters and translators, historians, flight attendants, sales representatives, authors and writers, CNC programmers, customer service representatives, telephone operators, travel agents and ticket sellers, and radio hosts and DJs.
- A proposed tipping point suggests that when AI can handle about 30% of a role’s tasks, productivity tends to improve, whereas much higher shares raise the likelihood of replacement.
- Jobs requiring manual skill, empathy or complex judgment—such as nursing, dentistry, construction and specialized technicians—show lower exposure to automation.
- The release sharpened calls for stronger governance and researcher training, with a Vatican-linked scientist urging ethical red lines and experts warning about bias, black-box models, reproducibility and fraud, while separate media use of ChatGPT yielded advisory degree lists not grounded in empirical study.