Contact us

Artificial Intelligence doesn’t exist. But it’s still gonna take your job.

news and insights
#Artificial intelligence

“They took our jobs!” This catchphrase is taken from an episode of the 2004 American series South Park in which American workers protest against men of the future immigrant in the present, to find work. It was originally a satyr against the alarmist rhetoric about the dangers of immigration. But this meme from 14 years ago is taking on a new lease of life today: artificial intelligence (AI) is coming from the future, and not a week goes by without a new speech explaining that it will replace us and take our work. Same alarmist talk, same passion in the rejection of the unknown as for migrants.



Fortunately, artificial intelligence does not exist. In French we say “dumb as a donkey”, and a donkey is in many ways much smarter than any computer. For example, we regularly talk about autonomous cars, which are not yet ready after more than half a century of research. Well, a donkey can drive itself only a few weeks after birth.

Another experiment: put a rat on Boston Dynamics premises, and the most sophisticated robot in a sewer. Which of the two will survive the longest? There is no debate. Even with the robot specifically programmed for sewer survival, it won’t last long. If intelligence is seen as the ability to achieve goals by adapting when the environment changes, then machines are still far from mammals.

In reality, the term artificial intelligence is vague for many. I borrow an analogy used many times in my professional environment (digital marketing): artificial intelligence is like sex for teenagers: everyone talks about it at recess, everyone says they know it, but nobody really knows what it is.

And like everything that is unclear, AI triggers fantasized and unreasoned fears-even if it means forgetting the less dramatic but more real dangers, and the benefits it can bring.


The spectrum of strong AI

The first of these fears is that of the “strong AI”, that is to say a conscious artificial intelligence, capable of thinking, and which would thus quickly supplant humans. Such technology does not exist today. But many people seriously (and publicly) claim that his arrival is imminent, first among them the singularitarians, including one of the apostles, Ray Kurzweil, predicted it for 2045. As a result, some people worry about Terminator or Matrix scenarios, and man-machine warfare.

These fears are not rational. There are still many unknowns in computer research that prevent any reliable prediction of the advent of a truly intelligent machine. Although enormous progress has been made to make computers more powerful, humans are still much better in many areas defining intelligence (planning in uncertainty, learning new skills with little or no explanation, ability to react to any situation or question…). These areas are currently what are called “open problems”: no one can say when a solution will be found. It is unlikely that this will be in the next two or three decades, given the efforts made so far, which have not been successful. Then, to say whether this will happen in fifty, one hundred or one thousand years is a prophecy to be left to the modern Nostradamus.

Let us draw a parallel: in the 18th century, people debated man’s ability to go to the moon (for example Fontenelle in his Entretiens sur la Pluralité des Mondes). Some thought it would never be possible, others thought it would happen in their lifetime. In the end, it took over 250 years. Who could have seriously predicted that at the time? Knowledge of physics and mechanics was not enough to know if and when we would go to the moon.

It’s the same with the strong AI today. Our knowledge of human intelligence is not sufficient to know when it will be exceeded. The pioneering researchers of artificial intelligence in the 1950s, Minsky and McCarthy, were convinced that it would take roughly 20 years for a machine to reproduce and surpass human intelligence, and they were seriously mistaken–when they were the greatest experts of the time.

It is in this sense that we can say that “AI does not exist”: the strong AI is not there, the machines do not know how to think, and nobody knows if, when, or how they can.

Transformation of work

But if A.I. doesn’t exist, what’s taking our jobs today?? These are new types of algorithms that–without thinking or will–can solve complex specific problems: recognize and sort images, load a warehouse, drive a car, play chess, etc. This is not a “strong” AI, because each algorithm is limited to its problem: the one who sorts the images cannot drive a car. However, the resolution of these problems was previously thought to be the prerogative of humans. Moreover, the computer technologies used reproduce learning and thinking behaviours that make them appear to “think” like men (although this is not really the case). This is why these algorithms are often called “weak” artificial intelligence as opposed to strong AI. More and more complex tasks can now be handled by these algorithms–to the extent that these tasks do not require “strong” intelligence.

The second of the great fears around A.I. is therefore that for all these types of tasks, men will be replaced. If there are fewer people needed to do the job, then there is less work and people end up unemployed! No. Many examples show the opposite; agriculture is the first and most significant. In the Middle Ages, 90% of the population had to work in the fields to feed everyone. Mechanization has completely changed the situation: much of the agricultural work has been done by machinery. There are now less than 2% of farmers in a country like France (although massively producer and exporter). The 88% of people who are no longer farmers have found a good job: there are more hospitals, more schools, and many jobs that did not exist before have developed.

The second example was the industrial revolution, during which modernization automated some of the workers’ trades. Many jobs have disappeared, but others have appeared. Working time has greatly decreased, averaging 8 hours/day today instead of 13 hours/day 200 years ago — while producing more and better goods and services for less strenuous work (think of coal mines).


The imagination and needs of humans being unlimited, there is always room for new services, and therefore always work. Around what the weak A.I. still cannot do, new professions will be created, and other already existing professions will be more widespread and better valued –creative or relational professions for example. The work will be shorter, more desirable and more interesting.


Safe Progress?

So is there nothing to fear from the I.A.? Can we rejoice without thinking? No.

Elon Musk, Bill Gates, the late Stephen Hawking, and the other technologists and researchers who have sounded the alarm, are obviously not crazy. There are real dangers posed by AI, but as with any technology, it is not its potential that is to be feared, but the men who control and develop it. Political dangers, when governments equip police officers with facial recognition glasses to immediately identify suspects (in China). Ethical dangers, when battle drones (in the United States) will have the possibility to decide to shoot to kill, or when the I.A. who will optimize the strategy of a company will advise to dismiss or to keep an employee. Economic dangers, when we see that the companies that master these technologies are accumulating more and more wealth and more and more power over their customers and users-we think of course of GAFAs (Google, Amazon, Facebook, Apple) and BATXs (Baidu, Alibaba, Tencent, Xiaomi)


These dangers are not insurmountable; Musk, Gates and others insist on ways to avoid them:

  • democratization of technology, to avoid control by a small number. This is the meaning of the OpenAI initiative;
  • and regulation, to avoid general loss of control. This has a symbolic significance on the part of people such as Musk, openly anti-regulation on any other subject.


However, these are real dangers, where the massive and lasting loss of jobs is not one for the moment. This does not mean that changes in the labour market will be painless: as in any evolution, there is a transitional period. It is a political and social issue that must be tackled to avoid those left behind. But the long-term record will be good for workers. We are already beginning to see it, as this anecdote from Nissa, employed in a warehouse in Amazon shows. The robots took his job; it was to stack 10kg containers on top of each other. Now she manages the team of robots who stack the containers for her: it’s less tiring, more interesting, and it doesn’t destroy her health.


Work does not disappear. These are the lilac punch jobs that disappear, and when you know how the song ends, you can only rejoice. A.I. is a formidable means which will allow the end of the alienation of work, the end of burn-outs, bor-outs and other brown-outs. Then yes, artificial intelligence will take our jobs–because we will leave them for others in which we will flourish much more.

This article was first published on the french media Les Echos 

published by
Philippe Rolet, Co-founder & group CTO

This website uses cookies for tracking and optimization purposes. By continuing to use the website you are giving consent to cookies being used. For information on cookies and how you can disable them, please see our Privacy and Cookie statements.

Yes, I agree