The evidence is more nuanced than the headlines

AI is changing work, but the strongest evidence does not support a simple story in which entire occupations disappear at once.

The International Labour Organization's 2025 global assessment found that one in four workers is in an occupation with some degree of exposure to generative AI. It also concluded that, because human input remains important, transformation of jobs is more likely than wholesale replacement in many cases.

A 2026 ILO review of emerging empirical evidence found real productivity gains in some settings, but also found that gains are uneven and have not automatically translated into higher measured output, earnings or employment.

What changes first?

The first changes tend to happen at task level.

Examples include:

  • drafting and first versions
  • summarising and extracting information
  • classification and routing
  • repetitive administrative work
  • research support
  • standard customer-response preparation
  • basic coding and analysis support

A role can therefore remain while the mix of tasks inside it changes significantly.

What becomes more valuable?

As routine cognitive work becomes easier to automate or accelerate, employers still need people who can:

  • exercise judgement
  • understand context
  • communicate clearly
  • build relationships
  • make accountable decisions
  • verify output
  • redesign processes
  • work across functions

The World Economic Forum's Future of Jobs Report 2025 also identifies AI and big data among the fastest-growing skills, while continuing to highlight analytical thinking, creative thinking, resilience, flexibility and leadership.

What employers should prepare for

1. Redesign roles at task level

Do not begin with "Which jobs can we remove?" Begin with "Which tasks should change, which should remain human, and which new responsibilities appear?"

2. Build AI literacy

Employees need enough understanding to use tools safely, assess quality, protect confidential information and know when human review matters.

3. Fix data and process problems

AI performs poorly when the process is undefined or the information is fragmented.

4. Put stronger controls around consequential work

Customer commitments, financial decisions, sensitive personal information and high-impact decisions need more review than internal drafting.

5. Measure actual value

Time saved is useful, but it is not the same as business value. Measure whether quality, turnaround, cost, customer experience or revenue actually improves.

A practical employer position

The useful question is not whether an organisation is "pro-AI" or "anti-AI".

The useful question is which work can be improved, what risks are introduced, and how employees should be prepared for the new operating model.

Sources used

  • International Labour Organization, Generative AI and Jobs: A 2025 Update: https://www.ilo.org/publications/generative-ai-and-jobs-2025-update
  • International Labour Organization, The Impact of GenAI on Jobs, Productivity and Work Organization, 2026: https://www.ilo.org/publications/impact-genai-jobs-productivity-and-work-organization-review-empirical
  • International Labour Organization, Changing Landscape of Skills in the Age of AI, 2026: https://www.ilo.org/publications/changing-landscape-skills-age-ai
  • World Economic Forum, Future of Jobs Report 2025: https://www.weforum.org/publications/the-future-of-jobs-report-2025/