A few years ago, at a dinner where the conversation drifted toward the most unusual and uncommon professions, I met Alejandro. While the rest of us shared the usual routines of our predictable jobs, he stated his occupation with the calm of someone describing any ordinary task: “My job is to stay informed about things.” Faced with the table’s curiosity, he explained the architecture of his “business”. Alejandro read international newspapers, philosophical essays and economic analyses, and also watched many films and read their reviews; he then distilled that whole whirlwind into short, digestible summaries. His clients, important figures in politics and big business, lacked the time to keep up, but their positions required them to display in public a level of knowledge and culture they did not really have. They paid him to simulate culture and depth. Alejandro was, in essence, a knowledge prosthesis, a creator of mirages of erudition for leaders with no time.
Somewhere in my mind I filed the anecdote away, probably in the drawer marked “the vanities of power”, until it suddenly resurfaced while I was listening to a talk on “possible futures” at an artificial intelligence seminar organized by the Chilean Venture Capital Association. The faces in the audience, packed with executives and representatives of venture capital funds, all wore the same expression, somewhere between fascination and horror, searching the screen for any clue or key that might give them some certainty about the uncertain future that technology is bringing our way.
What triggered my memory was not the data being projected, but the figure of the speaker. The keynote speaker was a well-known science fiction writer: Julio Rojas (a highly recommended author, by the way), who makes a living by constantly and conscientiously asking the question: “What would happen if…?” He not only uses this premise to structure his works (“El problema del fin del mundo”; seriously, read it), but also works as a consultant for companies that need precisely someone to imagine possible scenarios on their behalf.
I remembered Alejandro not so much because of the technical content of the presentation, but because of the underlying phenomenon and the invisible thread connecting both characters. The leaders of yesteryear paid Alejandro to process and summarize the present for them, in the same way they now use any AI tool; today’s executives pay creators like Rojas to imagine their future. In both scenarios, the world of management seeks to outsource its most intimate mental faculties. It is deeply paradoxical that, at a seminar dominated by the constant fear of being replaced by artificial intelligence, the ability to imagine emerges as an essential and irreducibly human quality.
There is a marked tendency today to attribute almost demiurgic powers to algorithms. However, once the phenomenon is stripped of its commercial packaging, we find what it really is: a powerful statistical model. Artificial intelligence does not conceive truths out of the void; it calculates probabilities. Its function is to find (very) appropriate answers to our questions, as long as that tokenized information already exists in its training model. The machine processes, synthesizes and recombines what humanity has already said, written or designed. It is an infinitely sophisticated rearview mirror, able to predict the next millimeter of the road, but unable to leap toward what has never been named.
This is where the real turning point is revealed. Artificial intelligence is an irreplaceable and prodigious tool for executing, planning and accelerating processes. In a hypothetical universe in which the pyramids of Egypt never existed, if we asked an AI to plan the logistics, optimize the materials, design the architecture and coordinate the construction, it would solve it in a tiny fraction of the time it took in antiquity. It would minimize costs, anticipate tectonic faults and allocate resources with surgical precision.
However, there is something the machine could never have achieved: conceiving the very idea of the pyramid in a world where pyramids do not exist. It could not have woven together anguish in the face of mortality, fascination with the cosmos and the desire for transcendence into a monumental, and useless, structure raised toward the sky. Before any algorithm begins calculating how to transport the stone or optimizing construction technologies, someone has to dare to look into nothingness and imagine the structure for the first time.
We live captive to the fear of being replaced by synthetic intelligences. But the lessons of Alejandro and Rojas converge on the same diagnosis. We may delegate news summaries, data processing, programming or the logistics of our projects. What we can never delegate, without extinguishing ourselves, is the radical act of imagining the useless. Artificial intelligence may build tomorrow’s pyramids in a matter of seconds, but before that happens, someone has to imagine them first.
Read the original column at https://opinion.cooperativa.cl/opinion/ciencia-y-tecnologia/la-inteligencia-artificial-puede-construirlo-todo-menos-la-primera-idea/2026-08-28/064842.html
