Generative AI is shaking up the workplace. From chatbots that draft emails in seconds to copilots that analyze data or summarize reports, tasks that once required people are now increasingly automated.
But not all jobs face the same level of exposure. Some roles, especially those involving language, writing, and research, are directly in AI’s firing line. Others, particularly hands-on or physical jobs, remain much harder to automate.
About the Study
In July 2025, Microsoft Research released a study called “Working with AI”, based on 200,000 real conversations with Bing Copilot. Researchers mapped out how closely AI capabilities overlap with human work across different occupations. The result? An “AI applicability score,” showing how much of a job’s daily tasks AI is already performing successfully. (Read more)
Top 25 Jobs Most Exposed to AI
Researchers calculated an AI applicability score for each occupation, a percentage that shows how much of a job’s tasks AI is already handling successfully (like writing, research, or answering questions).
Here’s what the data says.
| Rank | Occupation Group* | Job Title* | AI Applicability |
|---|---|---|---|
| 1 | Language & Communication | Interpreters, Translators, Court Reporters, Telephone Operators | 49–42% |
| 2 | Education & Research | Historians, Postsecondary Teachers (Business, Law, Social Sciences), Survey Researchers | 48–31% |
| 3 | Travel & Hospitality | Passenger Attendants, Ticket Agents, Travel Clerks, Concierges | 47–37% |
| 4 | Customer Interaction | Sales Representatives (Services), Customer Service Reps, Telemarketers | 46–40% |
| 5 | Writing & Media | Writers, Authors, Editors, Proofreaders, Journalists, Broadcasters | 45–37% |
| 6 | Technical & Programming | CNC Tool Programmers, Technical Writers, Library Technicians | 44–33% |
| 7 | Finance & Administration | Brokerage Clerks, Insurance Underwriters, Paralegals, Legal Secretaries | 41–34% |
| 8 | Policy & Science | Political Scientists, Mathematicians, Economists, Research Analysts | 39–31% |
| 9 | Marketing & PR | Public Relations Specialists, Advertising Sales Agents, Product Promoters | 36% |
| 10 | Admin & Office Support | Office Clerks, Records Specialists, Data Entry Keyers, Library Clerks | 35–32% |
| 11 | Legal Services | Legal Secretaries, Paralegals & Legal Assistants, Court Clerks | 35–33% |
| 12 | Religious Services | Clergy, Directors of Religious Activities, Chaplains | 32% |
| 13 | Healthcare Admin | Medical Records Specialists, Health Information Technicians | 32% |
| 14 | Education & Training | Training & Development Specialists, Career Counselors, Instructional Coordinators | 31% |
| 15 | Data & Information Mgmt | Survey Researchers, Market Research Analysts, Social Science Researchers | 31% |
| 16 | Media & Communication | Announcers, News Analysts, Broadcast News Analysts, Radio DJs | 30–28% |
| 17 | Legal & Court Services | Court Reporters, Hearing Officers, Judicial Law Clerks | 29–27% |
| 18 | Office Support | Executive Secretaries, Administrative Assistants, Receptionists | 28–26% |
| 19 | Financial Operations | Credit Analysts, Financial Clerks, Payroll Clerks | 27–25% |
| 20 | Business Services | Management Analysts, Compliance Officers, Policy Analysts | 25–23% |
| 21 | Marketing & Sales | Market Research Specialists, Sales Agents, Fundraising Specialists | 24–22% |
| 22 | Library & Education | Librarians, Library Technicians, Archivists | 23–21% |
| 23 | Information Management | Data Entry Operators, File Clerks, Records Managers | 22–20% |
| 24 | Clerical Support | Bookkeeping Clerks, Payroll Clerks, Office Clerks | 21–19% |
| 25 | Education Support | Teaching Assistants (non-STEM), Tutors, Education Program Coordinators | 20–18% |
Occupation Group* Groups of related jobs based on the U.S. O*NET-SOC classification system used in the Microsoft Research study. These represent clusters of occupations with similar tasks.
Job Titles* Examples only. These are sample roles from O*NET that fall within each occupation group. They illustrate the kinds of jobs most exposed to AI but are not an exhaustive list.
Key Takeaways from the Study
- Knowledge-heavy jobs are most exposed. Teachers, translators, and writers show some of the highest overlap with AI tasks.
- Clerical and office support is vulnerable. Roles like clerks and assistants involve repetitive, text-based tasks, prime territory for automation.
- Travel & customer-facing jobs are shifting. Ticketing, sales, and call-center roles already see AI replacing or assisting human workers.
- Manual and hands-on jobs remain safe… for now. Think electricians, nurses, and trades. Physical presence, judgment, and dexterity keep them insulated.
Conclusion: Humans + AI, Not Humans vs. AI
The study’s numbers are striking: nearly half of the tasks in language-heavy professions already overlap with AI capabilities. But instead of fearing mass replacement, the data suggest a different future, one where AI takes over routine, text-based, and repetitive tasks, while humans focus on strategy, empathy, judgment, and creativity.
For job seekers, this is both a warning and an opportunity. If you work in highly exposed roles like translation, teaching, or customer service, upskilling into areas of human strength like problem-solving, leadership, and emotional intelligence will be key. For others in low-exposure, people-first fields, the message is reassuring: AI may change how you work, but it won’t erase the need for you.