It is impossible to surf a tsunami. You may have trained a lot, own the best board on the market and have studied the wave’s curve in every detail, but the reality is that, once you ride, the consequences are unpredictable. That is the exact moment at which humanity stands regarding artificial intelligence: beginning to ride a wave whose crest seems infinite.
The speed at which this technology advances is astonishing. According to Epoch AI, training compute for AI models doubles every five months. That learning pace means that a model that began training five years and eight months ago has today multiplied its power and capacity 12,400 times. If that pace were applied to how fast a car can go, it would be as if in less than six years we could have developed race cars capable of traveling at 1.5 million km/h.
However, you can’t reach the destination faster just by running. The impact that AI and its applications are having on our world is of gigantic proportions. We are no longer debating whether this new technology will revolutionize everything, because that is obvious and happening, but whether the transformation will be total and whether we will have the capacity to adapt our economy, our labor model, our education model, our democratic institutions and, ultimately, our societies. And above all, whether we will be able to do it before the train derails.
“We are no longer talking about whether this new technology will revolutionize everything, for that is obvious that it is doing so, but about whether the transformation will be total”
In any case, it is important not to fall into doomsday trends and apocalyptic prophecies. The advantages of artificial intelligence are evident. AI enhances our ability to translate and understand, automates repetitive tasks, optimizes the forecasting of demand for renewable energies, detects pests, droughts or potential wildfires, improving the resilience of crops, reduces waste in corporate supply chains, democratizes access to information as never before, accelerates research in physics, astronomy, chemistry or biology, turns diagnostic medicine into a more precise and faster science—almost infallible—and reduces from years to months the discovery of molecules useful for developing new medicines.
If we focus only on the good, AI shows us a safer world, with less tedious jobs, improved healthcare and greater energy and industrial efficiency. Yet, every coin has its reverse. Humanity again looks over the edge, this time with more force, at the three challenges that previous industrial revolutions also presented to us: employment crisis (which always heralds major social transformations), crisis of the political model and the ultimate risk of self-destruction. Moreover, this new technology has accelerated a crisis that humanity had not yet experienced: a deep crisis of our attention. Our cognitive capacity, a hallmark of humanity in the face of other species, is strained and scattered by the competition of other intelligences. Perhaps it isn’t the moment to adopt Ludditist stances, but the power of the wave compels us to calibrate these threats carefully.
First, it is essential to focus on employment. Work is not only the main source of people’s income; it is also one of the main pillars of their identity and the greatest explanatory factor of social and geographic mobility. The first two industrial revolutions employed millions of Europeans, moving them from the countryside to cities and creating a new social class—the workers—that conditioned politics from the late 19th century through the entire 20th century.
Today, the trend is the opposite. The early-20th-century digital revolution automated numerous industrial processes and contributed to offshoring/outsourcing. Many European regions have been trying to adapt for years.
“This new technology has accelerated a crisis that until now humans had not experienced: a profound crisis of our attention”
The appearance of AI has not only accelerated that destruction of employment but has extended it to layers of society that were once thought untouchable. Lawyers, doctors, consultants and engineers watch with concern as their tasks are replaced. Their bosses still bring added value that is hard to substitute: from organizing, negotiating and taking risks to making connections; but the automation of production or the review of reports and the execution of repetitive processes is leaving many people behind.
The first dismissal is non-hiring. According to SurePoint, the hiring of first-year lawyers at the 200 largest US firms has been virtually stagnant for four years, while in that same period their accounts show record revenues. This situation poses three big problems: what do we do with people who cannot find work? how do we train senior workers who were not juniors? and how do we combat the inequality that a world where hiring is more linked to networks than to degrees will generate? There seem to be no ready answers yet.
Second, AI poses a major challenge to our classical institutions and their democratic expression. It is not only the enormous accumulation of data and power in the hands of the owners of their providers, often greater than that of many states. Rather, it is the novelty of the fact that AI’s efficiency is contesting space with emotional intelligence and decisions colored by human bias.
“The appearance of AI has accelerated that destruction of employment and extended it to layers of society that were thought untouchable”
Is a judicial decision made by a digital agent without personal biases more just? Whom do we attribute a war crime committed by an autonomous weapon? How constrained is a ministerial or corporate decision that scales up through the corporation without passing through human filters and is centered exclusively on probabilistic criteria? Again, the answers are not clear, especially when the geostrategic temptations of actors unconstrained by democracy condition others’ capacity to respond.
Third, this month the chief executive of Anthropic, Dario Amodei, stunned the world by proclaiming that we should “slow down the development of AI” given the risk that its power hides for global survival. The curious thing was how quickly his competitors agreed with him, something rare among the top leaders of OpenAI or xAI.
Do those who know AI best have reasons to believe that the apocalypse is near? The truth is that there are enough reasons to be wary of them, for their interest in slowing the development of new actors and consolidating their oligopoly could lie behind such strategies.
“Earlier nuclear weapons could destroy the world, but their activation required the action of one or more humans whose destruction was also guaranteed”
Yet, the level of autonomy of this technology is frightening. Nuclear weapons could destroy the world, but their activation required human action whose destruction was guaranteed. In the words of Pope Leo XIV, AI does not know human experience because it has neither suffered nor loved. And someone who does not feel, if given the nuclear codes, is unable to value anything other than statistical criteria to decide whether to activate them. In that situation, perhaps the best statistical response an AI agent might find to guarantee world peace is to destroy the world. Who knows.
Finally, AI presents a new challenge that neither the steam engine, the rural exodus, the railway, nor even the atomic bomb posed: a profound crisis of attention. It is true that this general decline in concentration isn’t entirely attributable to AI. The real problem lies with screens.
The ubiquity of smartphones to inform, communicate, project ourselves in society, guide us, pay, manage our finances or organize our time concentrates all of our attention in one place. Social networks and their personalized recommendation algorithms are a perpetual party. Who will pay attention to their teacher if their deepest interests circulate non-stop in the palm of their hand? Why wait for a friend on a bench to look around if I have a hundred notifications? And how will I listen to that friend when they arrive if those notifications keep jumping?
AI accelerates this attention collapse by generating a greater volume of content, and more personalized. Its ability to deliver solutions and rewards in seconds breaks the educational process and delegates thinking. Why develop reading or calculation comprehension if a machine already does it? Why learn to write or create? Or why solve complex puzzles if it’s easier to write a prompt?
That cognitive laziness grows among those who have known nothing else: AI-native or nearly AI-native students don’t know what it means to summarize, search for sources or take time to find solutions. The results of the latest PISA report are disheartening. Urgent measures are needed to restore educators’ protagonism and put cognition to work and strengthen before entering the real world, where the success in innovating has shifted toward those who can think beyond merely getting the answer to the problem.
“Urgent measures are needed to restore educators’ protagonism and put cognition to work and strengthen before exiting into the real world”
In short, many questions and few answers. However, what seems clear is that the real challenge of humanity should not focus exclusively on trying to guess how far this technology will go. The true challenge is to steer that development toward a better society. With AI at the service of humanity and conceived as a tool to make our lives easier.
Artificial intelligence, by definition, cannot be emotional, but by definition its creators, developers and users cannot escape emotions either. Let’s be clear about it, become aware, and do, as we have always done, that the experience of living once again overrides technology in favor of a better society for everyone. And if we do not achieve it, hopefully we will be able to cut the power before the tsunami carries us away.
