A coffee with… Yoshua Bengio | An uncertain future

What will our future look like?

Yoshua Bengio warns us from the start: we may be disappointed with his response. “I don’t have a crystal ball,” he said.

If we want to talk to him so much about the future and technological development, it is not because we suspect him of possessing a crystal ball. This is because the Quebec scientist is one of the most renowned experts in the world in artificial intelligence, which will undoubtedly – ​​correction: which will – revolutionize our lives. His work on deep learning won him the Turing Prize (the “Nobel Prize for Computing”) in 2018.

“Any serious scientist should tell you the same thing: that the future is uncertain. It is very difficult to say what will happen, because it depends on us collectively. In what direction do we decide to invest, what social standards do we set ourselves,” says Yoshua Bengio.

For this café-which-is-not-one (he doesn’t drink coffee, that’s good because neither do I), Yoshua Bengio welcomes us to his house near the University of Montreal, where he has been teaching since 1993. Very close to Mount Royal, where he walks every day, an outlet where he often finds his best ideas.

Three things immediately strike you when you chat with Yoshua Bengio. His intelligence. Its rigor. His humility. “Rigour and humility are necessary for the success of scientific research. It happens that researchers become big heads, ”he says.

This is why he avoids making spectacular (and risky) predictions about the future. “The future is much more uncertain than we think of as a linear trajectory,” he says.

In any case, the future will not look like a Hollywood sci-fi movie where humans have to fight for survival against intelligent robots, he thinks. “These films are often exaggerated. »

On a philosophical level, the scientist (and science fiction lover, everyone has their little contradictions) however liked the moral of the film. Her (with Joaquin Phoenix and Scarlett Johansson), where a man falls in love with intelligent software. Discloser: at the end of the film, the intelligent software leaves it… to go live with other intelligent software. ” They [les machines] leave humanity behind and do their own thing. That’s pretty much the most plausible scenario. [d’un futur avec des robots intelligents]. It’s like the relationship we have with ants. You don’t want to crush them, they don’t change anything for you. »

Does the recent prowess of ChatGPT, a chatbot with astonishing precision, impress you? You haven’t seen anything yet, warns Yoshua Bengio. “There are things a 4-year-old can do that ChatGPT goes wrong,” he says.

One day, Yoshua Bengio is convinced, we will create a robot more efficient than a human brain. A robot that will learn like a child – an old theory from British mathematician Alan Turing.

“There is no scientific reason why we would not be able to understand where our intelligence emerges with our brain, and therefore to build machines at least as capable as the human brain. It could take 10 years or 100 years. We are like mountaineers climbing a mountain without a map. We advance, we progress, but you never know what the next obstacles will be. »

Does the possibility that robots are smarter than humans scare you? You are normal. Yoshua Bengio also fears the harmful effects of artificial intelligence.

“My nightmare is that it will become easier and easier for a really screwed up person with power, like Vladimir Putin, to use technology in a way that is very destructive to humanity. The more powerful our tools are, the more this possibility increases. Our economic, social and political system was not designed for that. »

It would also be a mistake to “play Frankenstein”. “We don’t want computers to have to choose between their survival and ours. We can build machines whose purpose would be to serve us, and they would be happy about that. It’s more safe. »

This is why he is campaigning for an ethical artificial intelligence regulated by the precautionary principle. He also contributed to the adoption of the Montreal Declaration for Responsible Artificial Intelligence. “We decide together – if we are a democracy – what we want [faire avec l’intelligence artificielle] and what we don’t want. It’s okay to slow down the deployment of tech. Let’s think about it a bit before making very powerful tools available to anyone. »

Properly framed, artificial intelligence could lead to incredible advances in many areas. Among other things in medicine, an area where Yoshua Bengio does a lot of his research.

“Today, doctors are in a way in the Middle Ages. We throw arrows and see what works best on a large population. But that will change, because we will have the tools to understand how cells work. Artificial intelligence will create suitable drugs and treatments for a person’s cancer. »

Yoshua Bengio was born in Paris to Moroccan parents. After a summer stay with the grandparents in Montreal, the Bengio family moved to the Quebec metropolis when young Yoshua was 12 years old.

As teenagers, Yoshua and his brother Samy buy personal computers with their pocket money and have fun programming. Reminder: we are at the end of the 1970s.

The brothers will both make impressive careers in artificial intelligence. Samy made his mark in Silicon Valley: he is now head of artificial intelligence research at Apple, after having been one of the heads of artificial intelligence research at Google for years. Yoshua has made a career as an academic researcher and has become one of the world’s leading experts in artificial intelligence.

The Turing Prize winner also has other interests. A subject that concerns him enormously: climate change. “It’s an almost existential challenge for humanity,” says Yoshua Bengio.

Since the pandemic, he has not flown, gives his conferences in virtual mode, and is worried about our slowness to respond to the climate emergency.

“Reacting to climate change is basically quite simple: you just have to agree on the price of carbon,” he says. If everyone put the cost of carbon at $200 a tonne, we would solve the problem. […] Are people willing to pay four times as much for gasoline? No, not at the moment. But this is a short-term calculation. It’s a downside in the balance, but there are other worse downsides [si on ne fait rien]. »

For climate change as for artificial intelligence, society can not help but adapt, he believes. “We have something in us to find solutions. I am fundamentally optimistic. But not with rose-colored glasses. »