Will AI take my job? What the evidence actually says
A balanced look at AI and jobs: which kinds of work are most affected, what the research shows so far, what history suggests, and practical steps to protect your career.
First published · AI Digested
It's one of the most common questions people ask about AI, and one of the hardest to answer honestly. Some technology leaders predict huge numbers of white-collar jobs will disappear within a few years. Others point out that every past wave of automation ended up creating more jobs than it destroyed. Here is what we actually know.
This time, office work is in the firing line
Past automation mostly replaced physical, routine work: factory lines, farm labour, switchboard operators. Generative AI is different. It's best at tasks involving text, images, code, data and routine analysis: writing, summarising, translating, customer service, basic design, bookkeeping and programming.
A 2023 study by researchers at OpenAI and the University of Pennsylvania estimated that around 80% of US workers have at least some tasks that language models could affect, with higher-paid, more educated jobs more exposed than in past automation waves. The International Monetary Fund estimated in 2024 that around 40% of jobs worldwide, and about 60% in advanced economies, are exposed to AI.
"Exposed" doesn't mean "replaced". It means AI could change how some of that work is done.
Tasks, not jobs
Most jobs are bundles of many different tasks. AI tends to take over some of them, not the whole job. An accountant might hand AI the first draft of a report but still need to understand the client, spot problems and sign off the work.
What happens next depends on the business. If AI makes each worker much more productive, a company might:
- do more work with the same people,
- lower prices and grow, needing more staff,
- or do the same work with fewer people.
All three are happening in different places.
What the evidence shows so far
Productivity gains are real in some jobs. A widely cited study of over 5,000 customer support agents found an AI assistant increased the number of issues resolved per hour by about 14% on average, and by about a third for the newest, least experienced workers.
Some freelance work has shrunk. Studies of online freelance platforms found fewer jobs and lower earnings in writing and translation after ChatGPT's release.
Entry-level roles may be hit first. Several analyses since 2025, including a Stanford study of US payroll data, reported declining employment for younger workers in AI-exposed jobs such as software development and customer service, while older, more experienced workers were less affected. Economists are still debating how much is caused by AI and how much by other economic factors.
Mass unemployment hasn't happened, at least so far. But the technology is improving fast, and businesses are still early in adopting it.
What history suggests
When ATMs spread in the 1980s and 90s, people expected bank tellers to vanish. Instead, cheaper branches meant banks opened more of them, and teller numbers in the US held up for years, with the job shifting towards customer service and sales. Spreadsheets didn't end accountancy; they changed it.
But history also shows the transition can be painful. Workers in shrinking roles don't always move easily into new ones, and the benefits don't flow automatically to everyone. The speed of change matters a great deal, and AI may move faster than past technologies.
Which work is more and less exposed
More exposed: customer service, data entry, basic copywriting, translation, transcription, routine coding, paralegal research, bookkeeping, telemarketing, basic graphic design.
Less exposed: skilled trades, healthcare and care work, teaching young children, social work, emergency services, roles centred on trust and relationships, and senior roles with responsibility for decisions.
What you can do
- Learn the tools in your own field. People who use AI well are likely to be more valuable, not less. Start with our tutorials.
- Lean into judgement and relationships. Understanding what a client really needs, persuading, negotiating, caring and taking responsibility are harder to automate.
- Become the person who checks the AI. Deep expertise is what lets you spot when AI is wrong.
- Keep learning. AI itself is a good tutor. See learn anything faster with AI.
- Watch your industry. Notice which tasks are being automated and move towards the ones that aren't.
The honest answer
Nobody knows exactly how this will play out, and anyone who tells you with certainty is guessing. What's clear is that AI will change a lot of jobs, some faster than others, and that understanding the technology puts you in a much stronger position than ignoring it.
Keep going
Common questions
Which jobs are safest from AI?
Work that combines physical skill in unpredictable settings, close human relationships, or accountability for high-stakes decisions: think plumbers, electricians, nurses, care workers, teachers and managers. Even these will use AI tools.
Should I stop my child studying a subject that AI can do?
Probably not. Deep knowledge of a field is what lets you direct AI and catch its mistakes. Combining a subject with strong AI skills is likely to be valuable.
Will AI create new jobs?
Almost certainly, as past technologies did, though it's hard to predict which ones and how many. New roles already include AI trainers, evaluators and people who build AI into businesses.
Sources and further reading
- Brynjolfsson, Li & Raymond (2023). Generative AI at Work (NBER working paper)
- International Monetary Fund (2024). Gen-AI: Artificial Intelligence and the Future of Work
- Eloundou et al. (2023). GPTs are GPTs: An early look at the labor market impact potential of large language models
- Brynjolfsson, Chandar & Chen (2025). Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence
Spotted a mistake or something out of date? Tell us and we will fix it. Read how we research and write our articles.
Read next
How to write emails with AI (that still sound like you)
Use ChatGPT, Claude, Gemini or Copilot to write, reply to and improve emails: from difficult complaints to job applications. Step-by-step, with prompts to copy.
Updated 22 November 2026
How to summarise documents, PDFs and long reports with AI
Use AI to understand long documents fast: contracts, reports, letters, research papers and meeting notes. Step-by-step, with prompts that keep the summary accurate.
Updated 22 November 2026
How to use AI in a small business: 12 practical ways to save time
Practical, low-cost ways small businesses and sole traders can use AI today: marketing, customer emails, admin, bookkeeping, social media and more, with prompts and safeguards.
Updated 22 November 2026