For years, a relatively comfortable idea has been repeated that the professions most protected from automation would be precisely those that require the most training, accreditation and specialization. The logic seemed simple: if a job requires years of study and sophisticated knowledge, it would naturally be more difficult to replace it with machines.
However, recent studies, including one from Anthropic, are starting to point to a different picture. They show that many of the professions most exposed to automation come from the world of knowledge worksuch as programmers, financial analysts, medical records specialists, or customer support. This view was reinforced by Mehran Sahami, from Stanford University, who highlights that the impact of Artificial Intelligence (AI) is mainly on tasks within each profession and not just on professions as a whole.
Data from Goldman Sachs also indicates that administrative and legal areas have a high percentage of potentially automatable tasks, unlike sectors such as construction, maintenance or cleaning, which have less exposure. In parallel, the AI Jobs Barometerfrom PwC, says that the greater the exposure a profession has to AI, the faster the skills required of workers will evolve.
The result is a less visible but potentially more profound phenomenon: not only can the number of jobs change, but also the very nature of skills within each profession.
However, the most relevant signal may not only be found in the most exposed professions. It is the distance between what technology can already do and what organizations actually do with it.
The history of technological innovation shows that true disruption rarely happens when something becomes technically possible, but rather that transformation occurs when organizations redesign their processes around these new capabilities. This is where professional roles change, team structures change and the work economy begins to be truly redefined.
Everything indicates that we are not necessarily heading towards a scenario of massive disappearance of professions. What is emerging is a significant increase in productivity per worker. Each of us will be able to produce much more and some teams will become smaller. Because the traditional funnel for entering the job market may narrow, as part of the initial tasks begins to be absorbed by AI systems.
Here, an essential question arises: if a team becomes 25% more productive, thanks to AI, what will organizations do with this increase in productivity? Do they produce more? Do they reduce team sizes? Or do they cut working hours? The answer is not in technology, it is in human choices.
This perspective is related to a concept known as Turing Trap: the tendency to use new technologies primarily to replace human labor, rather than using them to extend human capabilities and create forms of value.
Economic history shows that this outcome is not inevitable. Technologies that automate tasks also create new activities, sectors and professions. In fact, much of the job growth in recent decades has resulted precisely from the creation of job types.
The real issue is therefore not just technological. It’s institutional. AI can amplify human talent or replace it, increase collective prosperity or concentrate even more value in the hands of a few.
None of these trajectories are predetermined. The future of work in the age of AI will not be decided by algorithms alone. It will be decided by human choices about how we want to use technology.
Leave a Reply