If you have opened a news feed in the last year, you have probably seen some version of the same headline: AI is coming for your job. It is a fair worry. People have bills to pay, families to support and a job market that was tough long before any chatbot showed up.
But the real story is more nuanced than the headline, and a lot more useful. Every big wave of technology has reshuffled work. Each time, the people who came out ahead were the ones who learned how the work was changing and got good at the new parts.
This article looks at what happened the last few times, what is genuinely different now, how to use AI safely to get more done, and which skills are worth your time.
We have been here before
The worry is not new. The Luddites raised the alarm about machines in the early 1800s, a US Secretary of Labor did it in the 1920s, and the economist Wassily Leontief did it in 1982. Each time, work changed shape rather than running out.
Farms to everything else. In 1900, about 40% of US jobs were on farms. Today it is under 2%, and the country still eats. Those workers moved into industries that barely existed then, such as health care, finance and electronics.
Spreadsheets. When the first electronic spreadsheet arrived in 1979, bookkeepers and accounting clerks took the hit. NPR's Planet Money counted about 400,000 fewer of those jobs in the US since 1980, and about 600,000 more jobs for accountants. Cheaper, faster number work meant people asked accountants for more.
ATMs. Machines were meant to finish off bank tellers. They did cut tellers per branch by about a third, but they also made branches cheaper to run, so banks opened roughly 40% more of them. US teller numbers roughly doubled in the 45 years to 2016, and the job shifted from counting cash to solving problems and building customer relationships.
Economists give two reasons this keeps happening. David Autor calls the first the O-ring principle: when machines make some steps in a job reliable, the human steps that remain matter more, as tellers' problem-solving did once ATMs handled the cash. The second is that we never get enough. More productivity leads to new products, new services and new industries, and those need people.
None of this was painless. Technology did eliminate some jobs, farm work most of all. The farm states of the early 1900s answered by keeping young people in school until 16, which economist David Autor calls one of the best investments the US made in that century. The pattern is consistent: tools take tasks, demand grows for what is left, and the people who build new skills move with it.
What feels different this time
Some things really are different. Earlier machines mostly took over physical work or fixed calculations. AI can draft, summarise, translate and write code, so it reaches into office work that used to feel safe.
The early evidence is mixed. Stanford researchers tracking US payroll data found no widespread job losses tied to AI. They did find that workers aged 22 to 25 in the most AI-exposed occupations now stand about 19% below where their employment would be had it kept pace with young workers in less-exposed jobs, mainly because fewer of them are being hired. Experienced workers show no such gap, and where AI assists people rather than replacing tasks, employment is flat or rising. The authors stress these are patterns, not proof that AI caused them.
The researchers also noticed what held up. Young workers lost ground in jobs that rely on codified knowledge, meaning knowledge that can be taught from textbooks or written procedures. Experienced workers in jobs that rely more on knowledge picked up through practice, mentorship and real situations saw employment rise.
Economists have seen a version of this before. In his talk, Autor described how employment in the US and much of the developed world grew at the high-wage and low-wage ends while the middle shrank, because routine clerical and production jobs follow rules that software can learn to run.
Looking ahead, the World Economic Forum surveyed more than 1,000 employers and expects 170 million new roles and 92 million displaced by 2030, a net gain of 78 million. The same employers say nearly 40% of the skills needed on the job will change. 77% plan to upskill their people, while 41% expect to cut headcount where AI automates tasks.
Most of this evidence comes from the US and large global employers, so read it as a sign of direction rather than a forecast for any one job here. The practical reading is that AI changes tasks before it removes jobs, and the people most exposed are those whose work is made up mostly of tasks it does well.
Use it safely, and use it to get ahead
The simplest way to stop feeling threatened by AI is to use it on low-risk work and learn what it can and cannot do.
Where it helps
Job hunting: tailor your CV to a job spec using experience you actually have, practise interview questions and research a company before you apply.
Writing: produce first drafts of emails, reports and proposals, then edit them until they sound like you.
Learning: have it explain a hard concept, quiz you or walk through a worked example.
Analysis: write spreadsheet formulas, tidy messy data and summarise long public documents.
Technical work: explain unfamiliar code, draft tests and suggest fixes that you then review.
Ground rules
Protect personal information. Keep ID numbers, contact details, other people's CVs and medical or financial records out of public AI tools. POPIA covers personal information about other people, and your employer may have stricter rules.
Protect your employer's data. Client details, source code, unpublished financials and contracts stay out unless your company has approved the tool.
Check before you trust. AI can sound confident and be wrong. Verify facts, figures, names and anything with consequences before you send or submit it.
Stay accountable. If your name is on the work, the mistakes are yours. Read everything you send.
Be honest on applications. Use AI to say true things better, never to invent experience or qualifications. Banks and other regulated employers often verify them.
Follow workplace policy. If there is no policy, ask your manager before you start using a tool for work.
A quick test before you paste. Ask yourself whether you would be comfortable emailing the same text to a stranger. If not, remove names, ID numbers and anything identifying, or keep it out of the tool. Describing the situation works just as well as pasting it: "A client is unhappy about a late invoice; draft a polite reply that offers a call" needs no email thread.
Check the settings. Some tools let you switch off chat history or stop your chats being used to improve the product. Find that setting before you paste anything you care about, and use the version your employer provides if there is one.
A CV example. Paste the public job ad and your own bullet points, and ask the tool where your wording differs from the ad. Then rewrite the lines yourself, and remove anything you could not back up in an interview.
Skills worth building
The World Economic Forum expects demand to grow fastest for technology skills such as AI, big data and cybersecurity, while human skills like analytical thinking, resilience, leadership and collaboration stay critical. The Stanford work points the same way: knowledge learned through practice held up better than knowledge that can be written down. So build both.
AI fluency. Treat it the way Excel was treated a generation ago. Learn to give clear instructions, check the output and fit it into how you already work.
Depth in your field. Know your industry, your clients and your rules well enough to tell when an answer is wrong. In financial services that includes risk, controls, POPIA and FICA.
Data literacy. Read numbers critically, spot errors and be comfortable with spreadsheets, SQL or dashboards.
Judgement. Decide, rather than just produce. Ask what an output assumes, what is missing and what happens if it is wrong.
Communication and collaboration. Explaining, persuading and working across teams still depend on trust between people.
Learning agility. Try a new tool every few months and keep a note of what worked.
In technical roles, expect more of the value to sit in design, code review, testing, security and data quality, and less in typing out routine code.
Start this week
You do not need a big plan. Four small steps are enough to begin.
Pick one weekly task. Choose something routine, such as a summary, a first draft or a spreadsheet formula, and try an AI tool on non-sensitive material. Note what you saved and what you had to fix.
Set your ground rules. Write down what you will never paste into a tool, and check your employer's policy.
Choose one skill from the list above. Give it 30 minutes a week, whether that is a short data course, a certification in your field or a project at work.
Keep proof. Add results to your CV and profile, for example "cut report preparation from three hours to one". Concrete results are easier for an employer to trust than general claims.
So, is AI coming for your job?
Parts of it, probably. The whole job, much less likely. History suggests the people who do well are those who pick up new tools early, use them carefully and keep deepening the skills that machines handle badly.
If you are looking for your next step, upload your CV to MaestroRecruit once and our AI will build your profile. Then set job alerts for the skills you are growing.