Artificial intelligence is not waiting for some far-off future. It is already helping people write emails, summarize reports, answer customer questions, explain policies, and pull useful details from piles of information that used to take humans hours to sort through.
That does not mean every worker should expect a pink slip from a chatbot. But a growing stack of research points to the same uncomfortable conclusion: the first big wave of AI disruption is hitting information work.
Microsoft Research analyzed about 200,000 anonymized Bing Copilot conversations and found that the most common work activities people used AI for involved gathering information and writing, while AI itself was often providing information, writing, teaching, and advising.
The 10 jobs facing pressure
The most exposed jobs are not always the ones people expect. According to Microsoft’s AI applicability ranking, the top 10 occupations include interpreters and translators, historians, writers and authors, service sales representatives, computer numerically controlled tool programmers, broadcast announcers and radio DJs, customer service representatives, telemarketers, political scientists, and mathematicians.
What links such different careers? For the most part, they all depend on language, information, explanation, persuasion, or structured problem solving. Essentially, AI can step into part of the workflow, whether it is drafting a text, preparing a customer response, translating a sentence, or helping organize research.
That is why other jobs close to the same danger zone include journalists, technical writers, proofreaders, editors, public relations specialists, data scientists, personal financial advisors, web developers, management analysts, and market research analysts.
Not all of them will disappear. Many may simply be reshaped, but reshaped can still feel dramatic when it happens to your own paycheck.
Degrees are not armor
This is the part that may surprise college graduates. A higher level of education does not automatically protect a worker from AI exposure.
In fact, Microsoft’s research found broad AI applicability across education levels, especially in jobs built around digital tools and information-heavy tasks.
That makes sense when you think about the average workday. A marketing analyst may spend hours turning scattered notes into a clean summary.
A financial advisor may prepare client materials. A language teacher may draft exercises. Those are exactly the kinds of tasks generative AI is getting better at doing quickly.
Still, AI applicability is not the same as job replacement. Microsoft’s paper warns that employment and wage effects will depend heavily on business decisions, not just technical capability.
In other words, the question is not only what AI can do, but what employers decide to hand over to it.
The work less exposed for now
Physical, hands-on jobs look safer from generative AI in the near term. Microsoft’s lowest-scoring occupation groups included construction and extraction, farming, fishing and forestry, health care support, building and grounds cleaning, transportation, and maintenance roles.
The reason is pretty simple. A chatbot can explain how to fix a roof, but it cannot climb the ladder and do the job. It can write a checklist for a nurse, but it cannot provide the human touch, spatial awareness, and real-world judgment that many care roles require.
That does not mean those jobs are immune forever. Robotics and AI-controlled machines could bring a second wave of disruption to warehouses, roads, farms, hospitals, and construction sites. But for now, generative AI is far more comfortable inside a laptop than on a muddy job site.
A global shift is already underway
McKinsey Global Institute has estimated that by 2030, between 75 million and 375 million workers may need to switch occupational categories, equal to roughly 3% to 14% of the global workforce. That is not a small workplace adjustment, that is a labor market turning a corner.
The International Labour Organization has reached a similar but more nuanced conclusion. Its 2025 global index found that one in four workers is in an occupation with some exposure to generative AI, while 3.3% of global employment falls into the highest exposure category.
Clerical occupations remain the most exposed, and the ILO says job transformation is the more likely outcome because most occupations still include tasks that require human input.
The impact also varies by country. In an earlier ILO analysis, high-income countries showed a 5.5% employment exposure to automation, compared with 0.4% in low-income countries.
The technology may be global, but its effects depend on infrastructure, skills, and how quickly companies actually adopt it.
What workers should watch
For workers, the safest move is not to panic. It is to pay attention. If a large part of your job involves turning information into words, answers, plans, summaries, lessons, scripts, or recommendations, AI is already knocking on your door.
That does not mean you need to abandon your career. It means the valuable worker of 2030 may be the person who can use AI well, check its errors, understand the client, protect sensitive information, and make judgment calls that a model cannot responsibly make on its own.
At the end of the day, AI may not erase the office, it may rearrange it. The people most ready for that shift will be the ones who stop asking only “Will AI replace me?” and start asking “Which parts of my work can AI take, and what should I get better at next?”
The study was published on arXiv.
