Imagine the scene:
An Andorran teacher, known for his seriousness, opens his phone on a winter evening. Friends have urgently sent him a video in which we see… his face, his voice, his tone. Except it is not him. In the video, he calmly states that “global warming is an invention.”
Everything is fake. But everything looks real.
The video was generated by an AI using a few minutes of audio taken from YouTube. It circulates on WhatsApp, TikTok, X. Students comment on it, parents are worried. And in just a few hours, the rumor has more weight than all the scientific explanations he has patiently given over the years.
This is what we might call “the power of the false.”
When the false outruns the true
Researchers working on AI, neuroscience, and social networks are observing the same thing: false information spreads faster, farther, and lasts longer than true information.
Why? Because what is false is often: more spectacular, easier to understand, less disappointing (it doesn’t contradict us), more addictive… and therefore more profitable for advertising.
On social media, beliefs, knowledge, rumors, and lies all travel through the same pipes. The algorithm is not looking for truth; it is looking for our attention. If it discovers that a conspiracy video about vaccines or about snowfall in Andorra keeps people watching longer than a rigorous weather analysis, it will push the conspiracy video.
And each of us ends up locked in a bubble of content that confirms us: the Andorran worried about his ski business, the young person obsessed with cryptocurrency, the consumer of “miracle” tips for saving energy at home…
When everyone answers “yes” or “no”… where we should say “I don’t know”
A poll in a mainstream media outlet asks the following question: “Is this drug / this vaccine effective against the disease?”
At that point in time, very few people could honestly know. There were not enough data, not enough clinical trials, not enough hindsight. The only intellectually honest answer was: “We don’t know yet.”
The poll results? Most people answer “yes,” another sizeable group answers “no.” And only a very small minority dares to say: “I don’t know.”
This reflex is worrying. We have become very quick to take a firm stand on questions for which we actually do not know the answer. It is no longer a taste for truth; it is a taste for opinion at the expense of knowledge.
The philosopher Nietzsche spoke of a “taste for truth”: the deep pleasure we feel when we understand something about the world, even if it is hard, even if it contradicts what we thought.
Today, that taste for truth is in competition with: seductive myths, ideological narratives, spectacular images, and easy promises.
By endlessly questioning everything on the same level, we lose the ability to distinguish between a meta-analysis that synthesizes hundreds of studies across many academic jurisdictions, and a well-edited TikTok thread. Each person then feels entitled to decide alone what is “true for them.” It is comfortable. But for a society, it is explosive.
When politics starts deciding scientific truth
There is a simple but essential difference:
- Politics debates what is just and unjust: what should we do? Whom should we prioritize?
- Science debates what is true and false: what do we observe? What do experiments show? What do the data indicate?
In science, progress comes through protocols, experiments, publications, and meta-analyses with titles that are not very sexy for the general public, written in professional jargon. In politics, we move forward through debates, values, and compromises.
The problem arises when the norms of political debate descend into the scientific realm.
When people start saying, for example:
- “The climate is not warming because that doesn’t suit me politically.”
- “This AI is not dangerous because it serves my economic interests.”
We are no longer discussing what the data show, but what we would like them to show.
In Andorra, we already see this through two very concrete examples:
- the gradual decline in snowfall at low altitudes,
- the increase in extreme weather events in our valleys.
These phenomena are measured, documented, and modeled. Yet it is still tempting to oppose them with opinions or short-term interests: a good ski season, a real-estate project, a tourism investment.
Our brains in an information tsunami
Our cognitive biases were shaped, over the long course of human history, to help us survive in a world poor in information: a few messages a day, a handful of village rumors, a local newspaper, a short RTVA bulletin.
Today, we are immersed in a permanent information tsunami: notifications, short videos, polls, fake quotes, spectacular rumors.
To cope with this flood, our brain recycles its old reflexes: it relies on intuition, on group authority, on the “common sense” of the neighbor. But where those reflexes once helped us make quick decisions with very little information, they become dangerous when information is overabundant, manipulated, and staged by algorithms whose only compass is our screen time.
What if the most fertile ideas were born between disciplines?
Physicist and philosopher Étienne Klein emphasizes another point: by sacralizing the boundaries between disciplines—physics, law, economics, philosophy, computer science—we lose opportunities to think and to acquire knowledge.
Modern physics has shown that: what we see—the Sun apparently moving around the Earth—is an illusion from our point of view, and even the idea that “the Earth orbits the Sun” is still a simplification: in reality, the Earth follows a geodesic (a “straight line” in space-time) in a space-time curved by the Sun’s mass.
It is precisely here that transdisciplinary approaches—combining science, philosophy, law, and economics—become essential, including for a small country like Andorra, caught between massive tourist flows and a digital transition that has become a necessity.
AI and energy: watts that make your head spin
Let us look at some numbers to grasp the scale of the issue.
- A human brain consumes around 20 watts while functioning.
- A giant AI model of the kind used for global conversational services can, under heavy load, require power on the order of a gigawatt (1,000,000,000 watts) at the level of the infrastructures needed.
To put this in perspective: if we add up the power of the adult brains of humanity—about 5.2 billion adults—we get roughly 104 gigawatts.
In other words: the total power consumption of all adult human brains is, in order of magnitude, comparable to what a handful of large AI compute farms could consume if they were used continuously—around 1,500 terawatts (1 T = 1,000,000,000,000 watts).
And behind these abstract numbers, there is something very concrete: data centers that heat up, water used to cool them, power lines that must be reinforced and electricity that must be produced, CO₂ emissions if the electricity is not 100% renewable.
The next time you catch yourself endlessly scrolling TikTok in the bathroom instead of reading a book, think about this: you are not just draining your phone battery. You are consuming a piece of a global electrical system that is already struggling to remain compatible with climate goals.
Aitek: an AI that also knows when to stop
Faced with this reality, there are several ways to design AI.
One approach is to centralize everything in gigantic models that are increasingly hungry for data and energy.
Another, more frugal and closer to the concrete needs of organizations, is to develop agentic, zero-code AI, like the Aitek platform.
The principle of Aitek:
- specialized AI agents (monitor, analyze, decide, execute),
- that activate when needed,
- and, crucially, that deactivate when they are not needed.
Inactive agents are put on hold, unnecessary processing is avoided. Energy consumption is no longer a “hidden line item” but a built-in dimension of system design: the right amount, at the right time.
What do we really want to preserve?
One simple but crucial question remains:
What do we need to change in order to preserve what we truly want to preserve?
We talk a lot about “energy transition,” “digital transformation,” “the AI revolution.” Everyone is in favor, in principle. But what do we absolutely want to keep?
- Our current level of energy comfort?
- Our lifestyle (travel, leisure, screen use)?
- The stability of the climate and ecosystems in our valleys?
- The quality of our public services, our healthcare system, our education?
Depending on what we choose as the invariant, the policies to be implemented will be very different. As long as we do not dare clarify this choice, we remain in a verbal consensus, in slogans, without a clear project.
Nostalgia for “the good old days” does not help. No one, knowing what we know now, would accept dental care today as it was in 1920. We might dream of an old Andorran village—but certainly not of its hospital.
The poet Paul Valéry wrote: “We advance into the future backwards.”
We keep our eyes fixed on the past, we are paralyzed by the urgencies of the present, and we struggle to build futures that are both desirable and lucid.
It is not our limits that need to be crossed, but the illusion of where we believe they lie.
We will not be able to abolish physical limits—climate, energy, resources. Nor will we be able to return to a world without AI, without networks, without science.
What we can do, however, is shift our imaginary limits:
- step out of the illusion that “more algorithms = automatically more progress”;
- step out of the idea that “everything is equivalent” between a scientific study and a clever video edit;
- step out of the laziness that makes us answer “yes” or “no” where we should be saying “I don’t know yet.”
For a small country like Andorra, this shift is an opportunity: we have the size of a laboratory, the proximity of key actors, the ability to test sober and intelligent solutions such as Aitek, to better educate people in a “taste for truth,” and to bring together our scientific, economic, and cultural strengths.
