The AI Arms Race Has Begun: Why We Need A NATO For Artificial Intelligence
Like with the atomic bomb, artificial intelligence will divide the world into the haves and the have-nots, French columnist Édouard Tétreau writes. To win the AI arms race, France and its allies need a new transatlantic partnership.

AI robots working
-Analysis-
PARIS — The artificial intelligence tool ChatGPT and its future competitors have started an epistemological and anthropological revolution. This super-powerful tool, a "metalanguage" that feeds on all the human knowledge available online, will disrupt every part of our lives.
We will think and make decisions differently with ChatGPT. We will perform better at work and be better educated, better fed and better supervised, collectively and individually. Whether in manufacturing, intellectual production or essential services like medicine — nothing will escape the power of ChatGPT and artificial intelligence.
Last month, The Wall Street Journal published a lengthy discussion of ChatGPT signed by academic Daniel Huttenlocher, former U.S. Secretary of State Henry Kissinger and Eric Schmidt, former boss of Google.
The authors ask the right, philosophical and essential question: that of trust. ChatGPT's answers have the appearance of intellectual and moral authority (drawing on all the world's online knowledge), but the answer is produced in a black box of machine-to-machine communications, which no one can enter.
The answers are therefore more likely to appear not as scientific, rational fact, but as religious truth. In ChatGPT we trust. Like a divinity that would take shape.
Like the post-Atomic era
I don't know how an atomic bomb works, but I understand its power enough to know that I want my country to have one, and not my neighbor. But ChatGPT is the Gutenberg printing press + the atomic bomb. A profound redistribution of world power will take place with ChatGPT and AI.
Like with the atomic bomb, there will be a club of nations with the means to exploit it.
As with the atomic bomb, there will be a club of nations with the means to exploit AI — those who will design the future of the world, and then those who will obey.
This arms race can be summed up in three figures: China has announced a goal of $150 billion in AI investments by 2030, aiming to achieve world leadership. Given the country's strategy of surprise developments in the fields of armament (e.g. balloons), this objective has likely already been reached and surpassed.
Transatlantic pact
The U.S. government, aware of the danger, is investing tens of billions per year in both the Pentagon and Silicon Valley. The European Union? Just one billion per year through the Horizon Europe and Digital Europe plans. And France? The "national strategy" for AI has propelled the rate of investment to 400 million per year.
Our future military, economic and social enslavement can be read in these figures. What should we do? Two things at the same time: first, build a transatlantic military AI pact to coordinate the efforts of allies — a NATO for AI.
Second, repeat the French feat of 1954, which allowed a country that had doubted itself since 1940 to acquire the atomic bomb in just a few years (from 1954 to 1960). This mustering of financial resources, engineers and extraordinary political will allow the country to hold onto its global standing and autonomy for decades.
Political will
AI is today the equivalent of nuclear power yesterday; there is no doubt. France has the financial resources: a country capable of burning €140 billion in three years to heal the wounds of the COVID-19 pandemic can allocate €50 billion for its very survival.
The same goes for human resources: we produce some of the best engineers in the world, and it is up to us to move them from professions of the past, like finance, and into those of the future.
Two questions remain unanswered: do we have an executive capable of carrying out this strategy? And do they have the real political will needed to get this done?