The 40 jobs ‘most at risk’ from AI – and 40 it can’t touch

Artificial intelligence is no longer a future concept — it’s already reshaping the way we work. From automation to generative tools, understanding which jobs are at risk from AI is becoming essential for workers, employers, and students planning their careers. arXiv+1

How these lists were made (short version)

Jobs at Risk From AI: What the Research Shows

Microsoft Research built an “AI applicability score” by looking at how people actually use generative-AI tools (Copilot/Bing Copilot) and mapping those tasks to standard occupational activities (using O*NET). Occupations whose typical tasks line up strongly with what AI already does well scored highest — and that’s how the “40 most exposed” list was created. Importantly, high exposure doesn’t automatically equal job losses: history shows technology often changes job content rather than wipes it out. Still, the score is a useful signal for which roles are most likely to be transformed first. arXiv+1

40 occupations Microsoft flags as most exposed to current AI (short list)

These are occupations where AI already matches many routine tasks — generation, summarisation, translation, classification, basic analysis, templated writing, and scripted communication. (This is a condensed version of Microsoft’s ranked list.) Tom’s Guide+1

1. Customer service representatives
   
2. Sales representatives (services)
   
3. Market research analysts
   
4. Management analysts / consultants
   
5. Data scientists
   
6. Public relations specialists
   
7. Technical writers
   
8. Editors, proofreaders
   
9. Writers, authors, journalists, news analysts
   
10. Translators & interpreters
    
11. Web developers (certain front-end / templated tasks)
    
12. Business analysts
    
13. Statistical assistants
    
14. Advertising & PR roles (copywriting, campaign text)
    
15. Broadcast announcers / radio DJs
    
16. Telephone operators & call centre clerks
    
17. Ticket agents / travel clerks
    
18. New accounts / clerical clerks
    
19. Telemarketers
    
20. Brokerage clerks / financial clerical roles
    
21. Models / product demonstrators (routine visual content)
    
22. Hosts / hostesses, concierges (scripted/reservation tasks)
    
23. Personal financial advisors (parts of advisory simpler tasks)
    
24. Library science & some academic research support roles
    
25. Business & technical teaching (content prep automation)
    
26. Historians, political scientists (drafting, summarising research)
    
27. Mathematicians (routine proofs, symbolic tasks)
    
28. Statistical clerks / assistants
    
29. Management support / admin staff with repetitive tasks
    30–40. (Other similar office, clerical, and routine knowledge-work roles). Tom’s Guide+1
    

Bottom line: roles dominated by text, templates, structured data, and repeatable decision-rules are the first to feel AI’s effects.

40 occupations that look comparatively safe (hands-on, bodily, or social skills dominate)

Microsoft’s analysis found many frontline, manual, and care-focused jobs show low AI applicability today — these are much harder for current AI to replicate because they rely on physical dexterity, embodied judgment, or deep human-to-human connection. Example safe occupations: Tom’s Guide

1. Plumbers, electricians, roofers
   
2. Construction laborers and trades (carpenters, masons, painters)
   
3. Cooks, short-order cooks, and food-prep workers
   
4. Nurses’ aides and many direct-care health roles (hands-on care)
   
5. Surgical aides, phlebotomists, massage therapists
   
6. Janitors, cleaners, groundskeepers
   
7. Heavy-equipment operators, excavating and loading machine operators
   
8. Commercial divers, dredge operators, logging equipment operators
   
9. Automotive glass installers, tire repairers & mechanics (certain complex diagnostics aside)
   
10. Pest control workers, refuse and recyclable material collectors
    
11. Packaging and filling machine operators (manual oversight)
    
12. Bridge/lock tenders, highway maintenance crews
    
13. Forest & conservation workers, agricultural sorters & graders
    
14. Parking lot attendants, delivery drivers (longer term: automation could change this)
    15–40. (Many other skilled trades, certain healthcare support roles, and physically rooted occupations). Tom’s Guide
    

Bottom line: if your job requires fine motor skills, unpredictable physical environments, care, persuasion, or in-person trust-building, AI is less likely to fully replace you soon — though it may still become a tool you use.

What the bigger reports say (context)

• The World Economic Forum’s Future of Jobs research emphasises that AI will transform job tasks and create new roles — employers expect both displacement and significant job creation, and they call for reskilling at scale. World Economic Forum
  
• HR-focused research (e.g., SHRM) recommends targeted skill investments and cautious workforce planning because “applicability” is not the same as immediate layoff risk. shrm.org
  

Practical takeaways — for workers and teams

1. Map your tasks, not just your job title. If >50% of your daily tasks are templateable (reporting, summarising, drafting), you have higher exposure. Start automating the boring stuff and learn the skills the AI can’t (complex judgement, relationship building, domain expertise). arXiv
   
2. Upskill into hybrid roles: AI oversight, prompt engineering, model validation, interpretability, and human-in-loop design are growing. Employers value people who can use AI responsibly. Microsoft
   
3. If you’re in a “safe” manual/care job: learn adjacent digital skills (scheduling, safety analytics, remote diagnostics) so you can benefit when tools arrive.
   
4. For managers: redesign jobs to combine AI’s throughput with uniquely human strengths — that’s where productivity gains (and new roles) appear. World Economic Forum
   

Sources & further reading

• Measuring the Occupational Implications of Generative AI — Microsoft Research / arXiv (AI applicability score and lists). arXiv
  
• Microsoft Research blog: “Further notes on our recent research on AI and occupations.” Microsoft
  
• Tom’s Guide summary of Microsoft’s 40/40 findings. Tom’s Guide
  
• World Economic Forum — Future of Jobs Report 2025 (jobs of the future and skills outlook). World Economic Forum
  
• SHRM brief: Jobs at risk — estimating near-term automation displacement risk. shrm.org