Microsoft names jobs least impacted by AI — blue-collar and healthcare roles top the list as communication jobs face highest risk
Microsoft analyzed 200,000 anonymized Bing Copilot conversations to score “AI applicability” across occupations, finding that physical, hands-on jobs in construction, manufacturing and healthcare support show the lowest applicability, while communication-heavy roles face the greatest exposure to LLMs.
- Hands-on trades and healthcare support show the lowest LLM applicability and thus are relatively insulated from generative-AI task alignment.
- Communication and knowledge jobs — interpreters, writers, sales and customer service — rank highest in AI applicability.
- Applicability ≠ displacement: Microsoft stresses the study measures where LLMs align with tasks, not where jobs will be replaced.
- Local planning: San Luis Obispo leaders should expand trades training, healthcare certifications and AI literacy for customer-facing roles.
How Microsoft measured “AI applicability”
Approach: Researchers scored tasks by how well generative LLMs map to them, then mapped those task scores to occupations using O*NET. The analysis used 200,000 anonymized, privacy-scrubbed Bing Copilot conversations to infer which tasks users asked LLMs to perform and how well models handled them.
Read Microsoft’s notes and the full study: Microsoft research blog and the full paper (also available on arXiv).
What the data shows: hands-on jobs stay safer
Across occupations with the lowest AI applicability, shared traits include physical labor, unpredictable environments, and tasks requiring hands-on manipulation of equipment or materials. Microsoft’s reported top-10 least-impacted occupations include:
- Dredge operators
- Bridge and lock tenders
- Water treatment plant and system operators
- Foundry mold and coremakers
- Rail-track laying and maintenance equipment operators
- Pile driver operators
- Floor sanders and finishers
- Orderlies
- Motorboat operators
- Logging equipment operators
Fortune summarized Microsoft’s list: Fortune coverage.
The study also highlights manual trades and healthcare support roles — such as personal care aides and home health aides — as having low AI applicability because they require physical presence, interpersonal care, or dexterous tasks that current generative models cannot perform.
Jobs more exposed to generative AI: communication and knowledge work
Occupations that center on written and spoken language, research, or information processing register higher AI applicability. Microsoft’s high-impact examples include:
- Interpreters and translators
- Historians
- Sales representatives
- Writers and authors
- Customer service representatives
Why: These roles involve tasks where LLMs can generate text, draft messages, summarize information, or help craft responses, raising applicability scores. See the Fortune summary: Fortune coverage.
Researchers’ caution: applicability is not the same as displacement
“Applicability measures how well an LLM aligns with specific job tasks. It does not predict direct job loss.” — Microsoft researchers
Two key caveats highlighted by the team:
- LLM focus: The study measures where large language models can help; it does not analyze other automation like robotics, sensors, or industrial AI that could affect manual roles.
- Not a displacement forecast: Generative AI can augment many roles and boost productivity, but researchers do not claim LLMs fully perform whole occupations.
Microsoft’s methodology and caveats are available in the research blog and full paper (arXiv).
Implications for San Luis Obispo, California
Economic impact
San Luis Obispo County’s economy spans agriculture, construction, manufacturing, higher education and a growing health and tourism sector. Microsoft’s findings suggest local jobs tied to construction, farm operations, and hands-on maintenance may be among those least impacted by generative AI.
Employers in manufacturing and water or waste management should consider targeted training that pairs workers with AI for safety, scheduling and planning rather than replacing hands-on labor.
Workforce and education
Local community colleges and vocational programs — for equipment operators, welders and health aides — may see sustained or growing demand. Cal Poly and community education programs could expand certificate and apprenticeship offerings and add AI literacy modules for communication-focused programs.
See Microsoft WorkLab guidance: Microsoft WorkLab piece.
Public services and infrastructure
Critical local roles such as water treatment plant operators and bridge or lock tenders are listed among the least-impacted and remain essential. Public works departments should prioritize recruitment, training and succession planning while exploring AI-assisted monitoring to boost efficiency without replacing operators.
Social effects
Home health aides and personal care workers — noted as low on AI applicability — are vital for an aging population. Ensuring competitive pay and training addresses equity and care needs. Conversely, tourism and hospitality staff who perform guest communications should receive AI-literacy training so they can manage tools that handle bookings and messaging.
Limitations and open questions
- Dataset and scope: The study uses 200,000 Copilot conversations and reflects how users chose to interact with one product; it may not capture all real-world task use.
- LLM focus: The analysis centers on large language models and omits robotics, industrial control systems and narrow automation that could change risk profiles for some blue-collar jobs.
- Geographic variation: Occupational tasks differ by region, employer and regulation; applicability in one place may not generalize elsewhere.
For full methodology and caveats, consult the full research and the arXiv paper.
Practical next steps for local leaders
- Map local jobs to Microsoft’s applicability categories and identify where targeted training is most needed.
- Expand apprenticeships and certifications in trades and healthcare support.
- Introduce short AI-literacy courses for communications, sales and customer-facing staff.
- Use Microsoft’s research as a reference for planning: research paper and research notes.
Bottom line: The analysis highlights where LLMs align with tasks — privileging augmentation and local planning — and suggests SLO-area leaders should invest in trades, healthcare training and AI literacy to prepare for an evolving labor market.
