Artificial intelligence does not have to replace an entire profession to change it: automating some of its tasks is enough. It can draft a first version, summarise documents or classify data; human work can shift from carrying out these operations to reviewing them, deciding what needs to be done and taking responsibility for the results.
This distinction helps explain why the effect of AI on employment cannot be summed up as “it will destroy jobs” or “it will create more jobs”. The outcome will depend on which tasks can be automated, how companies adopt the technology and who captures the productivity gains.
AI affects tasks, not always entire professions
An occupation combines different activities. An administrative job, for example, may involve data entry, assisting people, coordination and handling exceptions. Generative AI—systems that create text, images or other content from instructions—can perform some of these activities more quickly, but that does not mean it can take on all of them or that a company will decide to eliminate the job.
The International Labour Organization (ILO) updated its analysis of occupational exposure to generative AI in 2025. It estimates that one in four workers worldwide is in an occupation with some degree of potential exposure, but concludes that jobs are more likely to be transformed than to disappear entirely. “Exposure” means that some tasks could be affected; it is not a prediction of how many jobs will be eliminated or when.
Administrative occupations remain among the most exposed, while advances in the technology’s capabilities in voice, image and video have also expanded the range of tasks that could potentially be automated in communication and content jobs. But theoretical exposure alone does not account for costs, errors, regulations, available data or the need for human oversight.
Why some jobs change more than others
AI is especially useful when tasks can be described in instructions, are repeated frequently and involve digital information. This can affect office jobs and skilled professions, not just manual or routine work. By contrast, activities that require a physical presence, direct interaction, professional responsibility or handling unpredictable situations may be more difficult to automate completely, although they too can incorporate AI tools.
Technical capability is only part of the story. A company has to decide whether integrating the tool is worthwhile, prepare data and processes, train staff and take on the risk of errors. Industry regulations and the confidence of customers and workers also play a role. That is why the availability of a technology does not automatically lead to widespread adoption or staff reductions.
Productivity, hiring and job quality
When AI takes over some repetitive tasks, it can enable a person to complete more work or spend more time on parts that require judgement. If this increases output or reduces costs, it can benefit the company and, in some cases, open up new activities. But increased productivity does not in itself guarantee better wages, fewer hours or more hiring: these outcomes depend on business decisions, competition and how the gains are shared.
The opposite can also happen: if a tool enables the same volume of work to be done with fewer staff, a company may slow down hiring or reorganise teams. The effect may first appear in tasks or vacancies, and not necessarily in immediate layoffs. At the same time, algorithmic management—the use of digital systems to assign, measure or monitor work—can intensify the pace or reduce autonomy if introduced without safeguards.
An OECD study published in 2024, based on job postings from ten countries, found changes in the skills sought for occupations most exposed to AI. Among the most common were management, business processes and social skills. The study also found indications that demand for some skills could decline in certain more exposed workplaces. These findings concern vacancies and demand for skills; they are not proof that AI has already caused a general fall in employment.
Who might benefit, and who might be left behind?
The effects will not be shared equally. The ability to benefit from AI depends on access to tools and training, the type of job and the opportunity to negotiate how processes change. The IMF has noted that occupations in advanced economies are more exposed, partly because they involve more cognitive tasks. This may create opportunities to increase productivity, but it also means a greater possibility of substitution in certain activities.
For young workers, the challenge is not necessarily to learn how to program AI models. The OECD observes that many people who work with these tools will not need specialised knowledge of artificial intelligence. Instead, it will be useful to know how to check their responses, protect information, communicate and understand one’s own sector well. These skills do not eliminate the risk of job displacement, but they can help people adapt when tasks change.
What we can say—and what remains uncertain
Current studies mainly measure potential exposure and changes in tasks or skills. They cannot determine with certainty how many jobs will disappear, which new ones will emerge or whether productivity gains will offset the jobs affected. Forecasts depend on how the tools evolve and on decisions that have not yet been made by companies or in public policy.
The practical question, then, is not only whether AI can perform a task, but who oversees it, who bears the consequences if it fails and how the benefits are shared when it saves time. Ongoing training, worker participation and clear rules for data use and oversight will be decisive in ensuring that change improves work rather than merely speeding it up.
Sources and references
- International Labour Organization, Generative AI and jobs: A 2025 update
- OECD, Artificial intelligence and the changing demand for skills in the labour market
- International Monetary Fund, AI Will Transform the Global Economy. Let’s Make Sure It Benefits Humanity
Image: , CSIRO | CC BY 3.0 | Wikimedia Commons.
Educational notice
This content is for educational and informational purposes. It does not constitute financial advice or an investment recommendation.