Microsoft study names jobs AI is least likely to replace — healthcare, trades and hands-on roles top the list
Microsoft Research’s analysis of 200,000 anonymized Bing Copilot chats calculates an AI applicability score, showing AI significantly reshapes tasks more than replaces jobs — with healthcare and hands-on trades least affected and communication-heavy roles most susceptible globally.
Key takeaways
- Method: Microsoft built an AI applicability score from 200,000 privacy-scrubbed Bing Copilot conversations combining coverage, completion rate and impact scope — the foundation of the study.
- High applicability: Communication, research and writing tasks show the greatest overlap with generative AI; text-driven office roles rank highest for AI impact.
- Low applicability: Healthcare, hands-on trades and roles requiring manual care, empathy or local presence score lowest for generative-AI applicability.
- Practical message: The study stresses transformation over replacement — AI complements and changes tasks; policies should focus on retraining, apprenticeships and targeted upskilling.
Main content
What Microsoft analyzed and how
Microsoft researchers pooled about 200,000 anonymized, privacy-scrubbed conversations between users and Bing Copilot to measure where generative AI aligns with real work. They calculated an AI applicability score that combines three elements: coverage (how often AI could be used), completion rate (how often AI can finish a task), and impact scope (share of job activities AI can assist).
The method and detailed results are described in the Microsoft Research paper and discussed in the Microsoft blog. Microsoft later published broader Copilot usage data in Copilot usage summaries.
Jobs most at risk and why
Microsoft highlighted roughly 40 occupations with the highest AI applicability scores — primarily knowledge and office roles dominated by text, data and routine communication. High applicability reflects strong alignment with generative-AI strengths: large language models excel at drafting, summarizing, translating and retrieving information.
- Interpreters and translators — top-ranked because translation and text tasks map closely to AI strengths. Source: Fortune.
- Writers and authors — content generation, editing and revision overlap with LLM capabilities. Source: Microsoft Research.
- Customer service and sales reps, telephone operators, ticket agents — scripted responses and information retrieval can be automated to a large degree. Source: Fortune.
- Postsecondary teachers (some fields), historians and research-heavy roles — AI can assist with lecture prep, summaries and literature reviews but not replace core judgment and teaching presence. Source: Microsoft Research.
Jobs AI is least likely to replace
The analysis does not publish a neat “top 20 safest jobs” list, but categories with low generative-AI applicability are clear:
- Healthcare roles that need hands-on care, clinical judgment, empathy and physical intervention — nurses, home health aides, therapists and many surgical roles. Source: Fortune and Microsoft Research.
- Hands-on trades and blue-collar jobs — plumbers, electricians, construction workers and many machine operators require physical dexterity and on-site judgment beyond current LLM capabilities. Source: Microsoft Research.
- Manual care and social-trust roles — home care, personal care aides and positions requiring local presence and strong human relationships remain in demand. Source: Fortune.
Note: Blue-collar jobs are not immune to other automation forms — e.g., truck driving faces vehicle automation and monitoring-system advances — but generative AI (LLMs) has limited overlap with many physical tasks.
What the findings mean for workers and employers
Core message: prepare for change, not sudden job loss. High applicability ≠ elimination; instead, expect task shifts where AI handles routine writing, summarizing or data lookup while humans retain judgment, care and hands-on work.
“No occupation is fully performable by AI alone.” — Microsoft blog notes on applicability vs. displacement.
Employers should plan how AI will alter job tasks and invest in retraining for higher-value work. Governments and community colleges should expand training in trades, healthcare and AI oversight roles — areas Microsoft and its Work Trend Index flag as growth opportunities. For employees, the path is to learn where AI complements their work — using tools to accelerate research, draft text or analyze data while preserving empathy, manual skill, complex judgment and local knowledge.
Implications for Paso Robles, California
Paso Robles’ local economy rests on agriculture, wine production, tourism and small businesses. Microsoft’s findings suggest local strengths in hands-on work make many jobs there less susceptible to generative-AI disruption — an advantage for community stability and job retention.
Economic impact
Vineyard workers, cellar technicians, hospitality staff, tradespeople and local healthcare providers are among those least likely to be disrupted by generative AI. That points to continued demand for skilled labor in farming, wine production and personal services — preserving local employment and on-site value creation.
Political consequences
Local leaders will face pressure to support vocational training and apprenticeships rather than one-size-fits-all four-year college pushes. Conservative priorities like local control, small-business support and investment in trade pipelines align with encouraging apprenticeships tied to wineries, construction and health services. See the Work Trend report for broader context.
Social effects
Expanding locally available training for nursing assistants, dental hygienists, electricians and plumbers can help retain workers and reduce out-migration, easing pressure on housing and local services by keeping jobs local and stable.
Cultural relevance
Paso Robles prizes hands-on craft — from winemaking to skilled trades. The Microsoft findings reinforce that cultural strengths can be economic strengths: promote local craftsmanship, farm-to-table businesses and tourism experiences centered on human hospitality, which AI cannot replace.
Practical applications for local leaders and employers
- Expand partnerships between wineries, community colleges and trade schools to create tailored apprenticeships and short credential programs for vineyard work, cellar operations and hospitality management.
- Invest in healthcare training programs for home health aides and nursing assistants to meet safety needs and job growth projections; Microsoft’s work highlights healthcare as a lower-applicability area for generative AI.
- Help small businesses adopt AI tools for booking and simple marketing drafts while protecting local jobs that rely on human contact; prioritize grants or tax credits for upskilling rather than replacement.
- Strengthen digital literacy so local workers can use AI safely and effectively as a tool, not a threat — drawing on Copilot usage patterns reported by Microsoft.
Sources and further reading
- Microsoft Research: Working with AI — Measuring the occupational implications of generative AI
- Microsoft Research blog: Applicability vs. job displacement
- Fortune coverage of Microsoft research
- FinalRoundAI: Jobs AI will replace first in 2025
- Microsoft Work Trend Index and related report
- Microsoft Copilot usage PDF
