A clerk hears that AI will take her job and starts planning around that. The future of work with AI is more specific than that: AI mostly changes tasks, and the effect on pay depends on how demand for the occupation shifts.
Workers, managers and policymakers each need a different reading of that answer. The wider rules landscape is covered in ai regulation explained. The sections below separate common misunderstandings from what the evidence supports, then connect job change to worker-protection policy.
Common misunderstandings about AI and jobs
The most frequent error is treating AI as a single event that either keeps or removes a job. The International Labour Organization (ILO) says in its analysis of AI’s impact on work that AI impacts tasks and will not eliminate most jobs entirely. A job is a bundle of tasks, so a tool that takes over one task changes the job without ending it.
The second error is assuming that a job that survives is a job that stays the same. The ILO notes that these task changes can lead to effects on wages, depending on how demand for the occupation shifts. A role can remain on the payroll while its pay and conditions move, so headcount alone is the wrong measure of success.
| Common belief | What the evidence supports | Planning use |
|---|---|---|
| AI will remove most jobs outright. | AI changes tasks and will not eliminate most jobs entirely. | Break each role into tasks before deciding what is at risk. |
| If the job survives, nothing changes for the worker. | Task changes can affect wages, depending on how demand for the occupation shifts. | Monitor pay and working conditions alongside employment numbers. |
| AI affects every group of workers equally. | Women are overrepresented in the most-exposed occupations, such as administrative jobs, and young labour market entrants face emerging negative effects. | Target redeployment and training at the groups most exposed. |
| Workers can ignore AI tools and wait. | Experts quoted by Harvard Business School warn that you might be replaced by a person who uses AI better than you. | Build practical fluency with the tools used in your own occupation. |
Figures and findings in the table come from the ILO’s work on AI and the future of work and from Harvard Business School’s discussion of whether AI will improve or eliminate jobs.
- Audit tasks, not titles. List what a role actually involves, then mark which tasks a tool could do, which it could assist and which it should not touch. This is more useful than debating whether the whole job is safe.
- Watch pay as well as headcount. If demand for an occupation weakens, wages can fall even when positions remain. Employers and unions should compare pay and conditions before and after a rollout.
Which jobs AI will affect
Exposure is uneven, and the strongest evidence describes who is exposed more than it gives sector-by-sector percentages. Treat any single exposure ranking with caution and look at the task mix inside each occupation.
Administrative occupations and gender
ILO research has shown that women are more susceptible to the automation effects of AI because they are overrepresented in the most-exposed occupations, such as administrative jobs. The exposure is a pattern in the workforce, not a personal failing. Employers with large administrative teams should therefore plan task redesign and redeployment first for those teams, and policymakers should check whether support schemes reach them.
Young entrants to the labour market
The ILO also reports emerging evidence of a more negative effect of AI on young labour market entrants, with the evidence it cites coming from the United States. That is a signal to watch, not a settled global pattern. Graduates and school leavers should expect entry-level tasks to be the ones tools handle first, which makes early exposure to AI-assisted work a practical priority.
Augmented roles and new work
Not every effect is a loss. A report summarised by the American Enterprise Institute describes AI as a tool with the potential to enhance human labor and create new forms of valuable work, while stressing that this outcome is not inevitable. That framing applies to generative AI as much as to other forms of automation: the same tool can deskill a job or strengthen it, so how employers deploy it and what rules surround it decide the result.
Skills that matter
The skill that matters most is the ability to use AI well inside your own occupation, because the realistic comparison is often with a colleague who has that ability. Fluency means framing a task for a tool, checking the output against what you know, and deciding when to accept, correct or reject it. The tool does not supply that judgement, so it stays with the worker.
Reskilling is therefore the practical counterpart to the exposure findings. The ILO lists digital skilling and upskilling among its policy priorities, aimed at supporting access to new technologies. For an individual, that points to short, task-specific learning tied to the work already being done, rather than abstract courses with no link to a job.
Workers who want a formal signal of responsible practice, particularly in roles that touch automated decisions, can compare credentials in AI Ethics Certification: How To Choose, Get Certified, And Stay Current In 2026 – Technopolicy. A certificate suits people moving into oversight or governance duties; for most roles, demonstrated skill with the tools in daily use will count for more.
Teams and management with AI
When AI enters a team, it also enters management. The ILO observes that algorithmic management tools, which are getting supercharged by AI, are used for recruitment, allocating tasks, monitoring and evaluating workers. That widens the question from what AI can do to who decides how work is assigned and judged.
For team leaders, the practical step is to know which of those four functions an AI tool touches, and to keep a named person accountable for each. For workers, it is worth asking how a tool influences task allocation or evaluation before it becomes routine. Clear review points, where a person can question or overrule an automated output, protect both quality and fairness.
Policies that protect workers
Worker protection works best when it follows the pattern of exposure. The ILO sets out policy areas that each answer a different part of the problem.
| Policy area | What it involves | Planning use |
|---|---|---|
| Cushioning negative impacts | Redeployment, social protection and active labour market policies. | Suits workers whose tasks are being displaced; pair automation with a route to other work. |
| Digital skilling and upskilling | Enhancing digital skilling and upskilling to support access to new technologies. | Suits workers who need practical fluency; link training to real tools and roles. |
| Stronger governance | Mechanisms to ensure rights are protected in the workplace, such as safeguarding against discriminatory algorithms. | Applies to hiring, task allocation, monitoring and evaluation tools. |
The policy areas above are drawn from the ILO’s article on rethinking AI’s impact on future work.
Read against the exposure findings, the mapping is direct. Concentrated exposure in administrative work and among young entrants calls for redeployment and skilling support. The spread of algorithmic management calls for governance, because discrimination in a hiring or evaluation tool harms workers regardless of whether their job survives. The unique risk is leaving each of these to a different institution, so that no one connects task change, wage effects and rights.
The debate also extends past workplace rules. Some ethicists question the pace of AI development itself, and their case is set out in Why Some Ethicists Oppose Accelerating AI: 3 Core Arguments and What They Mean (2026) – Technopolicy.
Frequently asked questions
| Question | Answer |
|---|---|
| Do new laws have to be written before workers are protected? | No. The ILO says the response should apply existing policies and regulations while adapting and developing new strategies and governance frameworks, shaped through social dialogue and framed by the Decent Work Agenda and international labour standards. |
| Is universal basic income part of the answer? | It is one proposal. Harvard Business School reports that experts Korinek and Altman have suggested new government policies, such as a guaranteed universal basic income, may be needed. The ILO’s own emphasis is on redeployment, social protection, skilling and governance. |
| Do small businesses need special support? | The ILO’s skilling priority includes helping small businesses close the digital divide, so reskilling is not meant to depend only on large employers. |
