• In this report published in April of this year by Funcas, it is stated that 21% of Spanish companies with more than ten employees already use AI.

• The same report also indicates that between 1.7 and 2.3 million jobs will be destroyed in the next ten years in our country as a consequence of the use of AI.

• This destruction will be offset, at least partially, by the emergence of new jobs, specifically, 1.61 million.

The report warns that the new occupations will affect different groups of workers and through different mechanisms, which means that many of the current jobs will not simply be replaced by more qualified ones, but will directly disappear.

For its part, in this report by the World Economic Forum of January 2026, it is indicated that 54% of those surveyed (some 10,000 global executives) believe that AI will displace a good number of currently existing jobs.

The impact of AI on the labor market has already arrived. And it is here to stay, expand, and deepen.

Nowadays, papers in English can be corrected with a generative AI program, which not only indicates when one has made a mistake, but also makes suggestions to the user to improve the expression of an idea. Just a couple of years ago, it was necessary to hire an editor to at least take a look at works written by non-native English speakers; today, no one turns to language editors or proofreaders in the research field. On the other hand, current AI systems, although still in a rudimentary way, can build paragraphs that resemble in content, style, and scientific orientation those written by academics, after properly feeding the system through the prompts introduced into the machine. Knowledge is increasingly aimed at generating true experts in the art of the prompt: whoever feeds the machine better, in a more precise and specific way, will obtain the best performance from it.

From another perspective, there are already at least four companies in our country that offer AI systems for legal assistance to prepare lawsuits before courts of justice. These systems indicate the case law to be cited in each case and suggest judicial strategies. It is true that all these systems require (for the moment) human supervision: the High Court of Justice of the Canary Islands fined a lawyer in February 2026 (only 420 euros) for making no less than 48 citations of case law that ultimately turned out to be false. However, it is evident that the impact of AI on the legal industry will be devastating in the coming years. And in the academic field, as well.

What to do about it? In this 2023 KPMG report, it is indicated that 61% of people do not trust AI. The figure has evolved slightly in favor of trust: in its 2025 report on the same issue, KPMG indicates that around 55% of people do not trust AI (which represents a gain of six percentage points compared to the previous report). That is, it is possible that as AI rolls out, trust in it will increase. However, the distrust figure remains very significant, and countries that use AI heavily, such as Finland, are the ones that distrust it the most.

In reality, the fact that people generally distrust AI is good news, in my opinion. And we should not convince them otherwise, because that distrust probably acts as a cushion to buffer the effect of AI on the labor market. The argument is as follows: if the current levels of distrust towards AI were maintained (which, as I say, I hope they are), the only way to combat it would be by placing a human behind the machine. This would protect our labor market significantly, or to put it another way, it would protect humans in the labor market.

This is, by the way, the obligation established by the Artificial Intelligence Act regarding high-risk AI systems. Every high-risk AI system must be supervised by a human. And since the definition that the AI Act makes of high-risk AI systems is so broad (for example, those that could potentially affect fundamental rights are high-risk), we can conclude that many systems, if not all, should be subject to this obligation.

Placing a human behind a machine is not simply a protectionist regulatory trick (which it is: in Italy they have been protecting small businesses for years, and this is accepted as a matter that is not only economic and political, but also cultural), nor is it a message about the duties ahead for Spanish and global trade unions. It is, furthermore, a positive legal requirement for AI to develop in the most ethical and moral way possible. Let’s put a human behind the machine: our labor market will thank us; and probably AI will, too.

Source: Antonio Estella, Professor of Global and European Economic Governance at Carlos III University