The Agriculture Hype
We are going through a strange period.
We are still hunter-gatherers. But for some centuries now, the world has been changing little by little: tribes are domesticating animals, cultivating fields; communities are gathering around these activities; trades are being born.
It is still only the beginning of the beginning, but you can follow the dotted lines and guess which way the wind is blowing. Of course, predicting is hard, especially when it comes to the future, but we know this much at least: the world will be completely transformed. We try to get this across to our fellow hunter-gatherers. Surely they see the same thing we do.
“Agriculture? Nothing but hype!”
Some have arguments to explain why progress will stop dead. Not enough knowledge (data); not enough water (for the terabeasts); not enough raw material; not enough energy…
Others have no opinion, but are just fed up with hearing about agriculture.
“In 1,000 years, we’ll still be filling our baskets with berries and hunting mammoth. Mark my words!”
There is no good French translation, as far as I know, for ai-pilled. La piqûre de l’IA, the AI jab? That isn’t it at all. To have swallowed the pill is to be convinced of the extreme power of the revolution. If, in retrospect, we realize that we exaggerated this power, we will be able to say that the pill was a charlatan’s elixir. If, on the contrary, the power proves real, it will have been a drug that helps us see reality as it is, like the red pill in The Matrix.
Still, we have to admit that so far, the spectacular progress in AI has trampled the claims of the blue-pill gobblers (Gary Marcus, Ed Zitron) and confirmed the intuitions of the red-pill swallowers.
Sometimes the difference comes down to one’s vision of time.
I am discussing the Artificial Revolution with a fellow interpreter, about my age.
“Bah, anyway, I won’t be around when it happens.”
“You have terminal cancer?”
We carry over into the present a temporality that no longer fits, that no longer fits at all, the contemporary. Gone are the days when changes came along at a leisurely pace. The Agricultural Revolution took some five millennia to take hold on a global scale; the Industrial Revolution, about a century and a half, a pace of change more than thirty times faster. If we apply the same multiplier to the comparison between the industrial and AI, we get roughly… 5 years. A bit short, I admit, but it almost fits the forecasting scenario AI 2027: if we don’t slow down, if we follow Trump’s order—“Let ’er rip!”—it is likely that everything will be decided by around 2029 or 2030, seven or eight years after the public launch of ChatGPT, or five or six years after the release, in 2024 and 2025, of the first powerful reflective and agentic models (OpenAI’s o1, Anthropic’s Claude Opus 4).
In such a scenario, the risks of human extinction are very high. According to AI 2027, that is the most plausible outcome; the happy ending of the “slowdown” branch, judged less likely, rests in part on a stroke of luck, by the authors’ own account—a successful alignment process, in just a few months, thanks to the chain of thought.
In the AI 2040 scenario—which is not a prediction, this one, but a recommendation—the Artificial Revolution is “stretched.” Instead of running at full throttle over a compressed period, it would extend over fifteen years, until 2040. This temporality is less organic, more “artificial,” precisely: it has to be imposed by political decisions. It will not come about on its own.
Such is the watershed. The vision of time: that is what makes the difference between going on living as if nothing were happening and feeling an immense vertigo. (It may be no bad thing, perhaps, to go on as before: I sometimes envy those spared by AI anxiety.)
A vertigo and an urgency: for want of being able to change the world, at least, to bear witness to the transition.
Nakhon Ratchasima, Thailand