The rapid expansion of artificial intelligence is sparking widespread fears, and not just among the general public—some of the very innovators driving AI forward are sounding the alarms too, in particular regarding employment. Mustafa Suleyman, a leading figure in AI, recently declared, “AI is fundamentally a tool to replace human labor.” Is this cause for concern? Not necessarily.
Mustafa Suleyman is not just anybody: he co-founded DeepMind, one of the pioneers of AI, which has since been acquired by Google. He worked there as head of applied AI. He is currently the head of Inflection AI, which he co-founded in 2022. When he talks about AI, you have to listen. But when he talks about the threat of AI replacing jobs, he’s talking about the economy, and you need not listen.
There’s no doubt that AI will have a profound impact on the job market. Study after study has been published on the subject. Some predict a cataclysm. The World Economic Forum estimates that AI will eliminate 90 million jobs by 2023. Goldman Sachs puts the number at 300 million. It was against this backdrop that Suleyman issued his warning. And yet, if history is anything to go by, we can qualify these predictions.
Just one example: When Edison invented the phonograph in 1877, he wanted to provide businessmen with a discreet means of communication. It would allow them to record and send their own messages on a wax disk, without having to dictate them to a secretary to avoid indiscretions. But it was not as successful as had been hoped. On the other hand, the phonograph was perfect for listening to music, and that’s how it developed, much to the dismay of Edison, who was scandalized that his invention was being used for something so vulgar. His invention revolutionized the way people listened to music. Not surprisingly, it led to fears of job losses and protests against its use, and later that of radio, known as “canned music”. Both were accused of putting musicians out of work – it was a human labor replacing tool. Already! And yet, these two inventions led to the creation of a gigantic industry that today represents over $140 billion in revenue and 1.9 million jobs in the U.S. alone.
Let’s do a little exercise. Imagine we’re in the 10th century. A new technology slowly emerges: the plow. The inventor warns: “The plow is a tool that will massively replace human labor”. Yes, that’s the idea. We are now in the 50s. Replace “AI” with “tractor” and you get “The tractor is a tool that will massively replace human labor. That’s the idea, again. From the shovel to the ox-drawn plow, and from the ox-drawn plow to the tractor-drawn plow, substitution is the key to progress in agriculture. The author of these lines remembers harvesting with his parents’ neighboring farmer in the late 70s: bales of hay were collected and piled on the cart by hand. It took about ten day laborers for three weeks. In the late ’80s, the round ball arrived, and the farmer harvested almost single-handedly, with the help of his son.
There’s nothing new about technology replacing human labor. Technology is knowledge accumulated in such a way that it can be used by someone who doesn’t have that knowledge. So substitution is its very purpose! To resist it would be like asking farmers to work with spoons to reduce productivity in the hope of preserving jobs. We’d be returning to a form of subsistence agriculture barely capable of feeding the population we knew until the Industrial Revolution.
The limits of technical expertise
While AI is a tool to replace human labor, it is no different than any other technology. Mustafa Suleyman thinks he’s discovered a new and dangerous phenomenon, when it’s been the principle of our economy for at least two centuries, if not millennia. He’s trying to scare us and doesn’t seem to realize that AI, like any technology, is a source of new activities and new jobs that are already visible today. His unfounded warning shows the limits of expertise: one can be a recognized expert in a technical or scientific field (in this case, AI) and not be competent on the economic and social consequences of the development of this field. To each his own, as it were.
🇫🇷 A version in French of this article is available here.
📬 If you enjoyed this article, don’t hesitate to subscribe to receive future articles via email (“I subscribe” in the upper right corner of the home page).
