0.2free~7 min

Will AI take more jobs or create more jobs?

1. The honest answer

Both. And the question framed that way hides what actually matters.

The data from the main labour-market research bodies points in the same direction. AI displaces some jobs. It creates others. The net number, on the central forecasts, is positive. The jobs created are not the same jobs as the ones displaced, and not always for the same people. That gap is what the headlines miss.

This unit walks through the numbers from three of the most-cited reports, then the asterisk.

2. What the World Economic Forum projects

The WEF Future of Jobs Report 2025 is the most recent large-scale survey of employers, drawing on responses covering more than a thousand global firms. Their headline estimate, across all causes (AI, demographics, energy transition, geopolitics):

  • 170 million new jobs created by 2030.
  • 92 million existing jobs displaced.
  • Net effect: +78 million jobs.

For AI specifically — broken out from the broader technology shift — the report estimates 11 million jobs created and 9 million displaced through 2030. A net positive number, but a smaller one than the headline.

The same report notes 40% of employers expect to reduce headcount in roles where AI can automate tasks, while almost half also plan to redeploy affected staff into adjacent roles rather than letting them go. Both are true at the same time.

The fastest-growing roles in percentage terms: Big Data Specialists, AI and Machine Learning Specialists, Fintech Engineers, Software and Application Developers, and roles tied to the green energy transition.

The fastest-declining roles: Postal Service Clerks, Bank Tellers, Data Entry Clerks. Routine, rules-based, no judgement.

3. What McKinsey projects

McKinsey's analysis of generative AI and the future of work in America frames the question by hours rather than jobs.

  • Without generative AI, automation could cover roughly 21.5% of US work hours by 2030.
  • With generative AI in the mix, that figure rises to roughly 29.5%.

The eight-point jump matters because of where it lands. McKinsey estimates that generative AI could automate 56% of work hours spent on "applying expertise" — planning, designing, analysing, drafting. For physical work — repairing machinery, delivering goods — the same jump adds only one to three percentage points.

The translation: knowledge work was the part of the economy that previously looked safest from automation. The current wave reverses the order.

4. What Goldman Sachs projects

Goldman Sachs's 2023 report on generative AI put the global exposure figure at the equivalent of 300 million full-time jobs worldwide. Roughly two-thirds of US occupations have some degree of AI exposure. For exposed roles, the share of work that could be automated falls between 25% and 50%.

The same report's optimistic scenario projects generative AI lifting global GDP by 7% over a decade as productivity gains flow through the economy. Goldman's own note: historically, technological waves that initially displace workers have over the long run created employment growth. That precedent does not guarantee anything; it sets a base rate.

5. The asterisk

The case for AI being net job-positive at the aggregate level is reasonably strong across all three reports. The asterisk is at the individual level.

The jobs created are concentrated in technology, AI, data, software, green transition. The jobs displaced are concentrated in clerical, administrative, data-entry, transactional work. Telling a 55-year-old data-entry clerk that an AI-Machine-Learning-Specialist role opened up in a different city does not move them into that role.

This is what economists call a transition problem. It is also what most workers actually experience. Aggregate numbers and individual outcomes are different things.

The other asterisk is that all three reports are forecasts. The WEF report is based on employer expectations. McKinsey's is based on task-level analysis and adoption-rate scenarios. Goldman's is based on occupational-content modelling. Forecasts revise. Read the dates on the reports above and assume the next revision is on its way.

6. What this means for you

The headline question — will AI take jobs or create jobs — is the wrong question for an individual. The right question is closer to: will the jobs being created include one for me, and is the road to it clear?

That question is what the rest of this course is about. The data above tells you the macro is roughly net positive. Your personal answer depends on which side of the transition you place yourself on, and the next unit is about whether that transition includes the role of developer specifically.

Sources cited above: