It’s quite classic, and common to all eras, to think that what’s happening to us is unique and that nothing so important has ever happened to anyone else; and in a way, it is true, because this is the only period we can be sure is real, because we’ve witnessed it first-hand.
It’s also quite common to see a technological innovation presented as a major revolution, changing the world more than any other before it.
Today, it’s also fashionable to say that our time is unique and that artificial intelligence will change history more than any other innovation before it.
Let’s start by putting things into perspective:
Many technologies changed the world long before this one: the mastery of fire, the lever, the wheel, the domestication of the horse, the use of iron and coal, printing, the steam engine, the internal combustion engine, the electric motor, the telephone; and many other apparently more modest technologies have brought about changes that are at least as important. Each time, these technologies have been used for better or worse, depending on the ethics of the societies that mastered them. There are few revolutionary technologies (such as painkillers or the contraceptive pill) for which it has been difficult to imagine clever uses, even if they do exist.
Today, a technology known as AI is considered revolutionary, revolutionising the art of forecasting. Prediction, based on data from the past, is nothing new: farmers, augurs and then meteorologists were the masters of it. Decision-making based on these predictions is nothing new either, and power has always belonged to those who could predict, or who knew how to put forecasters at their service. Like the generals who used the forecasts of augurs, or the financiers who used the forecasts of analysts. And today’s AI is merely the current culmination of a technical evolution that began more than a century ago and consists of quantifying the data that we used until now, more or less intuitively, to deduce laws from the past that enable us to forecast. AI is already incredibly useful, predicting breakdowns, changes in demand and supply, and changes in the behaviour of customers, suppliers, patients and students, within the narrow limits of what past data can allow us to analyse. It will continue to progress, predicting better and in a greater number of areas.
However, it will remain incapable of predicting major disruptions, and in particular of predicting what the next major disruption will be. It is also incapable, like all other innovations before it, of preventing humans from misusing it: and it is not because it usurps the name “intelligence” that it can provide this guarantee: there are so many perverts, villains and barbarians among those described as “intelligent”.
AI, like many innovations, is like a hammer: it can be used to build, or to destroy.
The worst is not certain, but it is possible. In particular, it may be found at the junction of the digital and the biological, when robots that create new molecules are coupled with artificial intelligences that imagine new molecules, or even genetic manipulations, and create these new living entities without human intervention, mixing animal, plant and human cells to create more efficient beings, without moral barriers, without conscience.
Is this impossible? Let’s not forget what La Fontaine wrote in the last line of an all-too-forgotten fable, “The Swallow and the Birds”: “We don’t believe in evil until it has come”.
For the time being, the key to the proper use of this technology, like its predecessors, is to check in each case whether it is being used to serve the economy of death, or the economy of life, and life in general. An AI that simply helped to extract more coal and oil, to design more addictive drugs, foods and games, would clearly be serving the economy of death. An AI that will help to cure better, educate better, manage water scarcity better, produce renewable resources better, make democracy work better, would be at the service of the life economy and therefore essential to protecting the future.
To judge this, we need to establish transparency at every stage, to have competent checks and balances, and not leave the decision to the Panglossian technologists, or to the companies that employ them, or to the financial markets that profit from them.
Perhaps we should even dream that the next revolutionary technology, after artificial intelligence, will be the one that makes it possible to deploy human benevolence.
Painting: Peter Paul Rubens, Decius Mus questions the auspices, 1617