Microsoft forecasts these jobs will remain secure from AI
AI may not replace jobs like embalmers, but knowledge work occupations could see greater AI integration.

Microsoft has released a study analyzing over 200,000 anonymized Copilot chats to determine which jobs are most and least affected by generative AI. The research, published in July 2025, revealed that AI excels at tasks involving information gathering, summarization, customer interaction, and written communication. This puts roles such as translators, technical writers, journalists, salespeople, and customer service representatives at higher risk of AI disruption, with task overlap reaching as high as 98% in some professions.
On the other end of the spectrum, jobs requiring physical presence, manual labor, or intimate human interaction show very low overlap with AI capabilities. These include nursing assistants, massage therapists, phlebotomists, construction workers, dishwashers, and various transportation-related roles. These professions remain difficult for AI to replicate due to their reliance on dexterity, situational awareness, or human empathy.
Importantly, Microsoft emphasizes that their findings concern task overlap, not full job replacement. A job being AI-exposed doesn’t mean it will vanish; rather, certain responsibilities may be assisted or optimized by AI. The shift may lead to redefined roles rather than mass unemployment, particularly in white-collar sectors where AI acts as a productivity tool rather than a total substitute.
The report also hints at the potential for AI to enhance creativity and strategic thinking when paired with human workers, especially in mid- to high-skill roles. Microsoft’s internal use of Copilot already shows professionals using AI for brainstorming, report drafting, and data insights, rather than full automation of decision-making or judgment-heavy tasks.
Ultimately, the study reinforces the idea that jobs relying on human touch, physical context, or non-codified judgment remain safe—for now. As AI evolves, however, continuous learning and adaptation will be crucial across all sectors to keep pace with hybrid human-machine workflows.
Sources: Business Insider, Axios, Futurism, Tom’s Guide