Microsoft’s research division has published a paper analysing 200,000 anonymised conversations with its Bing Copilot chatbot to gauge how readily generative artificial-intelligence tools can take on specific workplace tasks. The study assigns an “AI applicability score” to different occupations and identifies 40 roles where the technology can perform a large share of day-to-day duties. The most exposed positions are largely white-collar service jobs that rely heavily on language and information processing, including interpreters, writers, customer-service representatives, data scientists and journalists. By contrast, manual and technical trades—from dredge operators and water-treatment-plant operators to roofers—rank among the 40 roles least susceptible to near-term automation by AI. While the authors emphasise that AI is more likely to reshape work than eliminate entire positions, the findings add to concerns about workforce disruption. Education specialists and recruiters cited in industry commentary say universities and employers will need to accelerate training in hard sciences, research-and-development and advanced AI skills to avoid sharp placement declines forecast for graduating classes in 2027-28.