The AI revolution: We are debating curtains while the building changes

Compiled with the help of artificial intelligence, based on Spanish-language information. Source below the article.

I get the feeling we are arguing about the colour of the curtains while someone is changing the entire building. We worry about whether a pupil uses ChatGPT to do their homework, whether an AI has copied an illustrator too closely, or where to place the label that says “artificially generated content”. All of these are respectable, enormously important, hugely urgent issues and, probably, within ten years, fairly irrelevant.

It is endearing, really. It is as if in 1908 we had organised an international congress to decide whether cars could park in front of the stables. I am not saying all these debates are useless. Some are necessary. We are not facing a new app, nor another smarter phone, nor Internet 3.0. We are entering something far more brutal: the possibility of manufacturing intelligence and, what is more, putting it inside machines capable of acting in the physical world.

From muscle to mind: a revolution of a different order

The Industrial Revolution multiplied our muscles. The internet multiplied our capacity to communicate. Artificial intelligence can multiply our capacity to think, create, decide, research and produce. I do not know about you, but that strikes me as slightly more important than the metaverse. HAL 9000 is already here. Skynet is a couple of news bulletins away. And before anyone calls the fact-checking department: no, I am not saying that on Tuesday a naked Terminator will turn up in the village square asking for Sarah Connor.

I am saying something more prosaic and, surely, far more momentous: machines are starting to do things that, until five minutes ago, we swore were exclusively human. They told us a machine would never write. It writes. That it would never create images. It creates them. That it would never program. It programs. That it would not reason. It reasons, or does something sufficiently similar that half of humanity spends the day arguing about whether we can call it reasoning.

And while we hold that fascinating metaphysical discussion, the models keep improving. What is unsettling is that we do not even fully understand how some of their capabilities emerge. Behaviours, strategies and ways of solving problems emerge from the training processes that nobody has written instruction by instruction. That emergent capacity is precisely part of its power and also part of its risk.

The robots are coming, and the bottleneck is no longer technological

But there is something even more amusing: the robots are arriving. For years we watched them dance, jump and perform pirouettes and we thought: what cute little monkeys. An 80-kilo Tamagotchi costing three million dollars. Now another phase is beginning. The problem with some humanoid robots is no longer simply getting them to lift a box. The problem is starting to be whether they fit down the corridor, where they recharge, what certification they need, how they work alongside humans and whether they turn a profit.

In other words: the bottleneck is starting to cease being exclusively technological and is becoming industrial. And that is where things get serious. Because revolutions do not begin when a machine performs an impressive pirouette on YouTube. They begin when a finance director opens a spreadsheet and discovers that the machine saves them money. The day a humanoid robot pays for itself in eighteen months, no science fiction film will be needed. A PowerPoint will do. And believe me: a PowerPoint with a good return on investment has destroyed more jobs than Terminator.

The future will not wait for our regulations

The curious thing is that we keep talking about the future as if it were a legislative proposal we could approve or reject. As if there were a window where we could say: “Artificial intelligence, come back tomorrow.” That is not going to happen. China is not going to stop developing robots because Europe is drafting the definitive regulation on how an algorithm should behave. The United States is not going to switch off its data centres until we finish deciding which university degree a prompt engineer should study. And no company is going to give up a technology that doubles its productivity because we consider that things were going splendidly in 2007.

We can regulate the future. In certain respects we must. What we cannot do is abolish it. There is a difference between fitting seatbelts in cars and banning cars to protect carriage manufacturers. And sometimes I get the feeling that a good part of the Western debate consists of gathering the best available experts to study how to save the carriage industry. Meanwhile, the horses have already seen the problem coming.

We are automating the capacity to automate

The real earthquake will not be that certain jobs disappear. That has happened in every technological revolution. What is genuinely new is that we are automating the capacity to automate. Think about it. A human worker learns a profession over years. An artificial intelligence can acquire a new capability and replicate itself a million times. A human lawyer can review a certain number of documents a day. A system can review thousands. A teacher can look after thirty pupils. An AI tutor can look after thirty million and explain fractions using football, Pokémon or Rosalía if that finally gets the child to understand them.

A programmer can write a certain amount of code. A company can deploy hundreds of agents working simultaneously. A scientist has twenty-four hours in a day and, except in particularly worrying academic cases, needs to sleep. A machine does not. When intelligence becomes cheap, everything whose price depends on intelligence starts to change. And there lies the part we forget when we get apocalyptic.

This could go extraordinarily well

This could go extraordinarily well. We can imagine medical assistants capable of helping detect diseases much earlier, personal tutors for any child, accelerated scientific research, new medicines, better materials, cheaper energy, ten-person companies producing what once required five hundred, universal translation, automated bureaucracy. Well, that last one may be science fiction. If an AI ever manages to submit a Spanish public subsidy application correctly first time without the missing Annex VII-B bis, then yes: we will have achieved superintelligence.

Naturally there will be destruction. There will be companies that disappear, workers who suffer, professions that shrink and entire sectors that will demand protection. There will be fraud, scams, digital rubbish, monopolies and industrial quantities of idiots using a brilliant technology to do stupid things at a speed never known before. We are human. Let us not ask for miracles either.

Serious risks, but fear is not an economic policy

And there are serious risks. The more capable these machines become, the more important it is to ensure we can control them. It is not about putting on a Silicon Valley T-shirt and shouting “disruption!” while cheerfully handing over the keys to the planet. But neither can we turn fear into an economic policy. Because the greatest risk may not be that the machines arrive. The greatest risk may be that they arrive, transform the world and we find ourselves still holding a round table on how to protect the previous world.

There will be countries that use AI and robotics to produce more, research more and create new industries. And there will be countries specialising in preventing anything from changing. Then the latter will set up a committee to investigate why they have become poorer. Some countries can take advantage of these technologies to create production, wages and new industries; others may try to preserve for too long an economy that is already changing.

Forget exact predictions: the direction is what matters

That is why exact predictions bore me a little. Five years? Ten? Fifteen? No idea. Anyone who gives an exact date for these things should accompany it with Saturday’s lottery numbers. Humanoid robots may disappoint. Certain AI improvements may slow down. Energy, technical or economic limits may appear. The history of technology is full of futurologists who promised flying cars and ended up giving us electric scooters abandoned on the pavement. But debating the date does not change the direction.

This is already happening. And when a technology improves simultaneously in capability, cost, autonomy and ease of use, the interesting question stops being whether it will arrive. The question is what we will be doing when it does. We are still preparing children for exams that a machine can solve in seconds. We are still designing universities as if scarce knowledge were locked away in a library. We are still legislating work as if the employment model of 2035 were necessarily going to resemble that of 1995.

We are still talking about productivity as if putting a person in front of a screen for eight hours were a law of nature established by Moses on a stone tablet. Perhaps we need to start questioning a few things. Because the world to come will not need us to work more. It will need us to know what to do with machines that work an enormous amount.

Why I am very optimistic

And on that point I am optimistic. Very optimistic. All the great human technologies began by provoking fear because all of them destroyed a part of the known world. Fire was also extremely dangerous. Electricity killed. Cars ran people over. The internet was going to finish off half the economy and, indeed, it finished off a part of it in order to create a much larger one.

Artificial intelligence will do damage. But it can also brutally multiply our capacities. For the first time we are building tools that help build tools, machines that help discover and systems capable of extending our intelligence. That can be terrifying. It can also be marvellous. Perhaps in twenty years we will look at 2026 the way we look today at the early years of the internet and ask ourselves how we could fail to see it coming. Except that will not be true. We saw it coming perfectly well. The problem is that we were too busy arguing about whether pupils could use ChatGPT to do their homework.

The future has already connected to the WiFi

The future is not going to ask permission. It is not going to wait for us to finish the report, publish the regulation and appoint a monitoring committee. It has already come in. It is sitting in the living room. It has connected to the WiFi. And it probably already knows the password. So we can devote ourselves to longing for the world that is leaving, or we can prepare for the one that is arriving. We can try to slow down every change, or we can try to be the ones who best know how to take advantage of it. We can look at this revolution exclusively with fear, or understand that we are probably facing the greatest expansion of human capabilities in our history.

I have few doubts. Wild years are coming. Uncomfortable. Bloody difficult. Confusing. At times dangerous. Surely unfair in many moments. But also extraordinary. The future is going to change almost everything. And I suspect it will be so brilliant that, when we finally lift our heads to look it in the face, we are going to need a good pair of sunglasses.

🧭 Explore the Canary Islands with us
📰 Read more Canary Islands news ☀️ Check Canary Islands weather 🏖️ Find the best beaches in the Canaries 📲 Join us on WhatsApp

Source

Scroll to Top