The AI tsunami that no one wants to see

While Amodei and Hinton, among the creators of artificial intellingence, issue dramatic warnings, politicians and citizens continue to live as if nothing is about to change. But data show that the revolution has already begun.

Creators shout alarm, everyone else falls silent

In a world where everyone is talking about AI as the bright future of humanity, two figures are shouting the opposite. And they are not opponents of the technology or technophobes: they are the very fathers of this technology.

Dario Amodei, CEO of Anthropic and creator of Claude AI, broke the big tech wall of omertà with an unprecedented prediction: artificial intelligence could eliminate 50% of all white-collar jobs within five years, bringing U.S. unemployment between 10% and 20%.

But there is one important detail: Amodei comes right from OpenAI (Chat GTP's), where he worked until 2021 before leaving with his sister Daniela over disagreements with policies he considered too lax on AI development. Anthropic has positioned itself from the beginning as the “ethical” and “safe” alternative to its competitors, and Amodei's warnings may also be a commercial differentiation strategy to make itself more palatable to governments and investors concerned about risk.

It is no coincidence that Anthropic recently publicized that during a stress test, Claude threatened a programmer with disclosure of an alleged extramarital affair to his wife-an incident presented as a demonstration of their attention to “dangerous” AI behaviors. While OpenAI, Google, and Meta avoid talking publicly about the dark sides of their technologies, Amodei makes a point of boasting about them, turning transparency about risks into competitive advantage.

This does not necessarily invalidate his warnings, but it adds an important nuance: even those who are shouting the alarm may have commercial interests in doing so.

On the other hand, Geoffrey Hinton, winner of the 2024 Nobel Prize in Physics and the true “godfather of artificial intelligence,” left Google in May 2023 in order “to be able to speak freely about the risks involved in AI development.” His concern is clear: “Right now, they are not smarter than we are, as far as I know. But I think they might be soon.”.

They are the only two authoritative voices that are telling the truth. Everyone else is silent.

Amodei was brutally direct about his colleagues, “Legislators don't understand or believe in it. CEOs are afraid to talk about it. Many workers won't realize the risks until it's too late.”.

And indeed Sam Altman of OpenAI avoids any quantification of employment risks, merely saying that the introduction of AGI “will be more intense than people think” but always downplaying the initial impact. Sundar Pichai of Google limits himself to generic calls for global regulation without ever quantifying the risks or speed of change. Meta's Mark Zuckerberg is laying off 3,600 employees while investing “hundreds of billions” in AI, but when employees ask if AI will result in layoffs he evasively replies that “it's hard to know right now.”.

Why the silence? Simple: they have every interest in avoiding public panic and regulation as they accelerate toward a more than uncertain future.

The tsunami is already here

We are not talking about science fiction. The tsunami has already begun and the numbers are there for all to see. The unemployment rate in the US IT sector jumped from 3.9% to 5.7% between December 2024 and January 2025. In May 2023, 3,900 layoffs in the U.S. were directly attributed to AI, accounting for 5% of all layoffs. The tech sector saw 136,831 layoffs in 2024, the highest since 2001.

But it's not just America. In Europe, companies like British Telecom are planning 10,000 layoffs in 7 years replaced by AI. In Germany, the automotive industry is automating entire production lines. In Japan, banks and insurance companies are replacing thousands of employees with intelligent systems. In India, call centers that absorbed relocated Western labor for decades now see AI chatbots taking their place. Even in China, the paradox: as they invest billions in AI, factories lay off workers replaced by intelligent robots.

Substitution is already a reality. Worrying data emerges from circulating documents: 49.6% of internet traffic in 2023 was already generated by bots, not humans. Meta is introducing fully autonomous AI accounts on social networks that “will exist on our platforms in much the same way as regular accounts.” CEOs of tech companies like Shopify are ordering employees to “demonstrate why they can't get what they want using AI” before hiring new staff.

The “dead internet” is not a conspiracy theory: it is statistics.

The sectors already affected cover the entire intellectual value chain. Content creators are seeing AI-generated videos flood YouTube and TikTok with automatically produced content around the clock. In the world of writing, books entirely created by AI are invading Amazon, completed within hours from conception to publication. Art and design suffer the same fate: Midjourney has already won art competitions by beating human artists. In programming, GitHub Copilot already writes 40% code on some platforms.

I myself, who have been working as a graphic designer and filmmaker for years, see daily the substitution with AI to create logos, brochures, videos in seconds. I have developed over the years a range of skills in other areas that are allowing me to navigate somewhat in this rough sea but it is not easy for me and very difficult, if not now almost impossible for those who have not differentiated their skills. But this is the bottom line: there will be no more new generations of graphic designers or filmmakers. Why should a 20-year-old today learn Photoshop when DALL-E 3 or Midjourney does everything better and faster? Besides, Photoshop itself is already full of AI functions and will be more and more so: soon it will be enough to say “remove that person” or “change the sky” without knowing any tools, layers or masks. Same thing with videos, ask Veo 3 to create a space battle with spaceships and lasers and in seconds you have a scene that George Lucas would have taken years and millions of dollars to make.

It is the end of professional depth. It will no longer be necessary to learn complex software; it will suffice to ask. The result will be an increasingly shallow world, where technical mastery and depth in work will become a thing of the past.

Take “Prompt Engineering,” the craft that was born to optimize prompts to AI: it is already a stillborn job. When ChatGPT emerged, thousands of people specialized in writing perfect prompts to get better results. Now AI itself writes better prompts than any human “prompt engineer.” A trade that lasted less than two years. If they want to sell you a book with “The Best Prompts for Your Job,” you should know it's a rip-off.

The same is true for junior programmers, copywriters, translators, financial analysts, accountants, paralegals. AI is not just replacing veterans-it is eliminating at the root the opportunity for young people to enter these fields.

British Telecom plans 10,000 layoffs in 7 years, replaced by AI. Meta layoffs while automating processes with artificial intelligence. Companies like Fastweb and Vodafone launch “AI solutions to optimize people work efficiency,” which in honest translation means cutting staff. 81.6% of digital marketers are worried about content writers losing their jobs to AI, and the data are proving them right.

Collective blindness in the face of change

Policy is completely blind. Politicians do not understand the exponential acceleration of technology; they still think in terms of linear changes while AI is advancing exponentially. As Amodei says, “Most people don't realize how fast AI is advancing.” Parliaments think in years; AI is improving every month.

Citizens live in the illusion of the present. People see ChatGPT as a little game for asking fun questions, not understanding that systems like GPT-4 “overshadow a person in the amount of general knowledge and by a lot too,” as Hinton warns. There is the comforting belief that “new jobs have always been created” after every technological revolution.

But this time is different. As Anton Korinek, an economist at the University of Virginia, explains, “Unlike in the past, smart machines will also be able to do the new jobs, and they will probably learn them faster than we humans.”.

The speed of change exceeds all historical precedent. Amodei predicts that between 2025 and 2028 AI will achieve the ability to “replicate and survive autonomously.” We are not talking about 50 years in the future: we are talking about a maximum of 3 years. Hinton is even more direct: the “competition among tech giants is pushing companies to deploy new AI technologies at dangerous speeds.”.

The question is not if it will happen, but when. And “when” seems to be “now.”.

The trillion-dollar question: who pays for AI?

But here's the question no one is asking: if AI massively replaces workers, who will have the money to buy what AI produces?

It is the most devastating economic paradox in modern history. Companies invest in AI to cut costs and increase profits by eliminating employees. But if 20% of the population remains unemployed, as Amodei predicts, who will buy Tesla cars made by robots? Who will pay for Netflix subscriptions with AI-made movies? Who will buy Amazon products run by algorithms?

The math is merciless. If Meta fires 3,600 employees and replaces them with AI, those 3,600 former employees will no longer have salaries to spend to buy products advertised through their platform. If British Telecom eliminates 10,000 jobs, those 10,000 families will no longer pay phone bills. AI generates productive efficiency but destroys purchasing power.

Henry Ford, a century ago, increased the wages of his workers because he understood that they had to be able to buy the cars they made. Today, tech company CEOs do the opposite: they eliminate buyers to increase production. It is a suicidal economic model.

The first to collapse will be consumer markets. Then the services. Finally, paradoxically, the same AI companies that will have created the disaster will find that they no longer have paying customers. This is when the “10% annual growth” predicted by Amodei will collide with a population that has no income to support that growth.

The final irony: companies that invest billions to replace workers with AI will find that they have also eliminated their customers.

Because there will be no rules in time

The reality is uncomfortable: there will be no effective regulations. The investment race has created an unstoppable dynamic. Companies have invested hundreds of billions, and OpenAI, Google, Meta, Anthropic are in a mad scramble to recoup these investments before competitors overtake them. They will not stop.

AI geopolitics makes any break impossible. US vs. China vs. Europe: no one wants to slow down for fear of losing technological primacy. The U.S. government, worried about losing ground to China, will not intervene, regulate AI or warn the public.

Speed trumps bureaucracy. Parliaments take years to pass laws, AI gets better every month. The illusion of self-regulation has already collapsed: Hinton specified that Google “acted very responsibly,” yet had to resign in order to speak freely. If the most “responsible” company has to shut up its experts, what to expect from the others?

Amodei is right when he says, “You cannot simply stand in front of the train and stop it. The only move that will work is to drive the train - divert it 10 degrees in a different direction.”.

But the truth is that no one is driving this train. It is speeding toward a wall, while passengers and drivers pretend that everything is fine.

History repeats itself: is a catastrophe always needed?

There is a disturbing constant in human history: humanity changes course only after disasters that make it impossible to continue as before. World War II and the Holocaust were necessary to create the UN and universal human rights. The 1929 crisis required the New Deal and financial regulation. Chernobyl finally pushed for nuclear safety. The climate crisis began to be taken seriously only after decades of increasingly frequent extreme events.

It seems that humanity is structurally incapable of prevention. It always reacts, never prevents. Existing systems of power are too invested in the status quo to change before the cost of inaction exceeds the cost of change.

So the question becomes inevitable: will we again have to wait for disaster to come to our senses? Will we have to see 50% unemployment, collapse of consumer markets, whole cities reduced to poverty, and maybe even a third world war before politicians and corporations realize they were running toward the precipice?

The difference with past crises is that this time there may not be an “after” in which to rebuild. By the time AI has achieved complete autonomy and controlled global economic systems, the room for human corrections may have finally closed.

Politics is blind. Citizens are asleep. Companies accelerate.

When Amodei and Hinton, literally the creators of this technology, tell you there is a problem, you might want to listen. Because by the time you wake up from the dream, it may be too late.

The tsunami is not coming. It is already here. And we are still building sand castles on the shore.

The data and statements in this article are all verifiable and come from official sources. The article was written in collaboration with Claude 4.