As companies like Amazon publicly announce AI-powered layoffs, employees are scrambling to understand which careers may soon disappear and be outsourced to technology.
A report from Microsoft researchers studying the impact of generative AI on jobs provides some clarity.
Translators, historians, and writers are among the occupations with the highest AI applicability scores, according to the report, which ranks occupations, meaning their job tasks most closely match AI’s current capabilities. Customer service and sales personnel, who account for about 5 million jobs in the United States, will also need to compete with AI.
Overall, the researchers say the jobs most at risk are those involved in knowledge work, such as those working in offices with computers, mathematics, or clerical work. Salespeople also rank high because they often share and explain information.
Microsoft said that high applicability does not automatically mean those jobs will necessarily be replaced by AI, but the list of roles quickly spread and experts deemed them “most at risk.” This comes as companies like IBM freeze thousands of new roles expected to be taken over by AI over the next five years, and Indeed said UK graduates face the worst job market since 2018 as employers pause hiring and leverage AI to cut costs.
Of course, there are some jobs that are unlikely to involve AI. Bridge and lock tenders. Additionally, water treatment plant and system operators are among the jobs with virtually no exposure to generative AI, as they are required to actually use the equipment.
Still, business leaders like NVIDIA CEO Jensen Huang say it’s best to embrace AI because every job will be affected by it in some way.
“Every job is going to be affected immediately, there’s no doubt about it,” Huang said at the Milken Institute World Congress in 2025. “You won’t lose your job to AI, but you will lose your job to someone using AI.”
A degree won’t save you from the AI-driven employment revolution
Many of the jobs likely to be quickly upended by AI, such as political scientists, journalists, and business analysts, are jobs that typically require a four-year degree to get a job. And as researchers point out, a degree, once considered a surefire path to career advancement, is no longer a safeguard against changing times.
“In terms of educational requirements, we found that occupations requiring a bachelor’s degree had greater applicability of AI than occupations with lower requirements,” the researchers said. We studied 200,000 real-world conversations from CoPilot users and cross-compared the AI’s performance against occupational data.
On the other hand, there are some career paths that have less exposure to AI and are in increasing demand. The medical field is particularly vulnerable to this problem. According to the U.S. Department of Labor, the home health and personal care assistance industry is expected to create the most new jobs over the next decade.
At the same time, the researchers recognized that even their findings did not capture the full scope of the AI revolution, and that further automation could occur through more than just generative techniques. “Our measurements are purely about LLM, and other applications of AI could certainly impact professions that involve operating or monitoring machinery, such as truck drivers.”
Kiran Tomlinson, a senior researcher at Microsoft, said: luck This research focuses on identifying how AI has the potential to change the way work is done, rather than replacing or replacing jobs.
“While our research shows that AI can support many tasks, especially those involving research, writing, and communication, it does not show that it can perform any single occupation perfectly. As the adoption of AI accelerates, it is important to continue research and better understand its social and economic impact,” Tomlinson said.
The big bet on Gen Z’s education may not be all that appealing.
After experiencing a roller coaster of layoffs across the tech industry in recent years, many Gen Zers have turned to seemingly stable fields like education.
The sector was the fastest growing industry among UK graduates last year, and has similarly become a top career choice for US graduates. And while this profession can offer more work-life balance and commensurate benefits, the ability of AI to do the job may pose further headaches. The report highlights agriculture and home management educators, as well as post-secondary economics, business, and library science teachers, as roles where AI is more applicable.
While it is unlikely that schools will adopt AI teachers en masse, the report’s findings highlight how quickly technology could reshape many professions, including the teaching profession.
Top 10 professions least affected by generative AI:
- dredger operator
- Bridge and lock tenders
- Water treatment plant and system operators
- Casting mold and core manufacturer
- Track laying and maintenance equipment operators
- pile driver operator
- floor sander and finisher
- the orderly
- motor boat operator
- logging equipment operator
Top 40 professions most affected by generative AI:
- interpreters and translators
- historian
- passenger crew
- service sales person
- writers and authors
- customer service representative
- cnc tool programmer
- telephone operator
- Ticket agents and travel agents
- Broadcast announcer and radio DJ
- Securities company clerk
- Farm and Home Management Educator
- telemarketer
- concierge
- political scientist
- news analyst, reporter, journalist
- mathematician
- technical writer
- Proofreader and copy marker
- host and hostess
- editor
- Business Teacher, Post-Secondary Education
- Public relations specialist
- Demonstrators and product promoters
- advertising sales agency
- new accounting clerk
- statistics assistant
- Counter and rental clerk
- data scientist
- personal financial advisor
- archivist
- Economics teacher, higher education
- web developer
- management analyst
- geographer
- model
- market research analyst
- public safety correspondent
- switchboard operator
- Library Science Teacher, Higher Education
A version of this story was first published on Fortune.com on July 31, 2025.
