9,500 years ago, in what is now central Turkey, a group of people decided to settle down. They put down roots, built walls and stored grain. What emerged from that decision was Çatalhöyük: it was not yet a city, but it was the seed of something never seen before. The fascinating thing is that those pioneers could not have known what they were building; they had no way of anticipating that their settlement would, millennia later, give rise to the alphabet, the printing press, the Industrial Revolution and the Internet.
Every great leap forward in human history shares this initial blindness: those living through it do not realise they are crossing a fundamental threshold. Their descendants understand it; that is, we. Jensen Huang, CEO of Nvidia, suggests that we are at a similar juncture; according to him, no one knows for certain if he is right, because Artificial General Intelligence (AGI) is not a switch that flips on suddenly, but a blurred threshold. History teaches us that world-changing moments — such as the birth of Jericho or the first cities between the Mediterranean and the Zagros Mountains — are rarely appreciated whilst they are happening. They are recognised afterwards, when there is no turning back.
That is why understanding the present is not a matter of technique, but of perspective.
Let’s take a virtual trip to Seoul, 12 March 2016. The second game of the historic showdown between machine and man. Lee Sedol, the world Go champion with 18 international titles, faces AlphaGo. Go is an ancient Chinese game of immense complexity (perhaps we could say the same of chess...); Sedol knows every corner of that board better than any living human being.
Suddenly, the unthinkable happens. On move 37, the machine places a stone in a position that nobody understands. In 2,500 years of Go history, no human would ever have played there. The commentators call it a mistake; according to the experts, it ‘makes no sense’. Sedol stands up, leaves the room and spends nine minutes outside. When he returns, his face is that of someone who has just seen something for which he has no words. 157 moves later, that ‘absurd’ position built a winning structure that no master could have anticipated.
To grasp the significance of this breakthrough, one must understand the scale involved: chess has around 10 to the 43rd power possible positions (which can be handled by computational power). Go has 10 to the 170th power—more positions than there are atoms in the observable universe. No machine could win by brute force; it needed what DeepMind’s engineers called ‘judgement’. AlphaGo was not programmed with human rules; it learnt on its own, playing against itself millions of times, until it built a representation of the game incomprehensible even to its creators.
That same logic leapt from the board to science with AlphaFold. In 2020, this system solved protein folding, a biological puzzle that had remained unsolved for over 50 years. Whilst the best human teams achieved accuracy scores of 40 out of 100, AlphaFold scored 93. In 2024, this breakthrough was recognised with the Nobel Prize in Chemistry.
But something else happened, and it is happening today: AI began to improve AI. AlphaTensor discovered matrix multiplication algorithms that are more efficient than those known since the 1960s – the very foundation of modern computing. Essentially, the tool began to sharpen the tool. This raises an uncomfortable question about control and the pace of what lies ahead.
Unlike the dot-com bubble of 2000, this time there is a real and disruptive product. Copilot reduces programming time to a third; diagnostic systems detect cancers with over 95% accuracy. But here the difference is not one of degree, but of nature.
Data is often compared to oil, but oil is consumed. Data, on the other hand, accumulates and is reused. This creates what the economist Daron Acemoglu describes as a trend towards extreme monopoly. More data produces better models; better models attract more users, who in turn generate more data. The cycle closes, leaving out any small competitors.
And what about jobs? Joseph Schumpeter spoke of ‘creative destruction’: the car put an end to the horse-drawn carriage, but created new industries. However, the current revolution is different. It does not target repetitive physical jobs, but rather mid- and high-level cognitive tasks: analysis, writing, diagnosis, programming. Sectors that were once the refuge of those escaping industrial automation. Goldman Sachs estimates that 300 million jobs could be affected; today, perhaps half of them have already been affected. The problem is not just how many jobs are disappearing, but who gains and who loses: the benefits are concentrated in capital, whilst the risk falls on the worker who must reinvent themselves without an educational system to support them.
The stakes are also geopolitical. In Taiwan, an island of just 36,000 km², 90% of the world’s most advanced chips are manufactured. Without TSMC, there would be no GPUs to train models, nor any infrastructure for the future. Before his death, Kissinger called Taiwan ‘the most dangerous place on the planet’; you can well imagine why...
Today, three models of governance are competing: the American model (libertarian, run by private boards of directors); the Chinese model (state-run, with big data and surveillance); and the European model (regulatory, seeking to ensure that AI serves the citizen). These are not merely domestic policies; they are projects vying to define humanity’s global standard.
Perhaps the most profound impact is not economic, but personal. What becomes of a society that can no longer tell whether what it reads was written by a human or by a persuasive algorithm? (We experience this every day at HIVE.) AI exploits our cognitive biases, producing content that ‘sounds’ true because it mimics our markers of credibility.
Matthew's school blackboard.
As Seneca wrote: “It is not because things are difficult that we do not dare; it is because we do not dare that they are difficult”. We lack the political courage to regulate a technology that is redefining who we are. Work is not just a salary; it is our identity and our way of connecting with others.
Years after his defeat, Lee Sedol said something startling about that 37th move: ‘I felt as though I was witnessing something that came from nowhere I knew... I felt genuinely alone on the board’. That loneliness is the realisation that something non-human has begun to create knowledge, and we are all caught in the eye of the storm.
AI is not merely a tool that extends our capabilities, like the steam engine. It is a technology that replicates and, in certain domains, surpasses our very essence. That is why the important questions are neither technical nor economic. It is not “When will AGI arrive?”, but “What kind of world do we want to build with it?” Will tomorrow be like the Terminator saga?
Silicon Valley doesn’t have the answer, nor does a committee of experts, nor does an algorithm. Move 37 in our story has yet to be made. It is not a technical move; it is ethical, political and cultural. And the board is waiting for us.
