When a system can perform some of our work in a few moments, the next question is about the time it gets rid of. Will it become room for better activity, increased tasks, or a reason to reduce staff? The answer also depends on the decisions of those who organize the work and the power of those who can participate in it.
Artificial intelligence makes this question particularly urgent because it intervenes on administrative, linguistic and professional activities. However, talking only about the capabilities of software risks hiding the structure within which it is introduced: property, infrastructure, contracts and the possibility of negotiating change.
A job is not a whole trade
The International Labour Organization updated its analysis of generative AI exposure in 2025. Estimates that one in four workers is in an occupation with some degree of exposure. He made it clear that these are potentialities and that, for many occupations, transformation remains more likely than complete replacement. ILO, research update.
The distinction is about what we measure. A trade includes different tasks and relationships with people, places, and responsibilities. The technical possibility of automating a part does not automatically indicate how many people will lose their jobs, when or with what effect on wages.
We can imagine the same tool in two organizations. In the first one it reduces repetitive activities and allows you to better follow difficult cases. In the second it is used only to increase the rhythms. It is an example of social reading: technological change acquires direction through choices that must be able to be discussed.
Behind the interface are infrastructure
The UNCTAD report of 2025 recalls the concentration of investment and inequalities in AI infrastructure. It also points out that in 2022 one hundred companies accounted for about 40% of the company-funded research and development in the world: a figure on overall industrial research, not an AI-limited count. UNCTAD, Technology and Innovation Report 2025.
This dimension helps to formulate a question about power. If tools, computing capabilities, and access to data depend on a few subjects, what alternatives remain available to workers, small organizations, and countries with fewer resources? Access to an interface does not necessarily coincide with the ability to determine its conditions.
Centralization can therefore be studied through concrete elements: who owns an infrastructure, who sets the price, who can change a rule and who bears the consequences of the interruption of a service. Criticism becomes more solid when it follows these relationships instead of relying on an indistinct image of power.
A transformation that includes those who work
We can propose a criterion: to evaluate automation also through the quality of the experience of those who use it. They count the time it takes to control a result, the ability to correct it, and the space that remains to learn. A faster process can transfer new loads of responsibility without making them immediately visible.
The comparison at work should therefore include people directly involved, stated goals and tools to observe the effects. Training can be thought of as an opportunity to participate: understanding what changes allows you to discuss it with greater autonomy.
The perspective of consciousness enters here in a concrete way. It invites us to ask what idea of a human being the organization is leading: a resource to compress or a presence capable of judgment, learning and relationship. The future of work also takes shape in the response we choose to give together.
