Incredible as it is, artificial intelligence as we know it today has the power to quickly become “more powerful than a tyrannosaur” and pose an existential threat to humans – hence the urgency of frame it without delay.
This is the message that Israeli historian and best-selling author Yuval Noah Harari and renowned Quebec researcher Yoshua Bengio delivered on Wednesday during a conference given to a packed room as part of the creative business event C2. Montreal at Place-Ville-Marie.
Both experts agreed that artificial intelligence (AI) has the potential to become an existential threat to humans.
“AI is different because it’s a decision-making technology,” said Harari, who was speaking by video conference from Israel. The flint knife or the atomic bomb, it takes a human to decide whether to use them. But the AI makes decisions, and therefore takes away some of our power. »
According to Mr. Bengio, a professor in the Department of Computer Science and Operational Research at the University of Montreal, AI threatens the next US presidential election in November 2024 in the short term.
Tools like ChatGPT can be used to deploy propaganda, disinformation, etc. campaigns. So it’s a threat to democracy. To lose democracy is to lose all the good things that liberal society has given us.
Yoshua Bengio
Within a few years, such technologies could have an intelligence equal to, and then superior to, that of humans, he said.
“We don’t know if that’s going to happen. But if so, it’s existential as a threat. Imagine a new species so intelligent that looks at us like we look at a frog… Do we treat frogs right? »
Mr. Harari has also expressed concern about this possibility. “Artificial intelligence is a caterpillar that can become a tyrannosaur. And it won’t take millions of years of evolution for that to happen: we could be there in a few years. »
The two men recently signed an open letter calling on all AI labs to immediately halt the training of more powerful AI systems for at least 6 months.
For Yuval Noah Harari, private companies will probably continue the search. “The most urgent thing is for the government to control the deployment. For example, we could have a simple rule: the AI cannot impersonate a human, we must know that we are talking to the AI. For thousands of years, humans have had laws against counterfeiting money, and pretty much those laws work. We need similar laws on human counterfeiting. »
Governments are newly interested in AI because of the popularity of tools like ChatGPT. But elected officials still misunderstand the full potential of this technology, including what it means for the future of democracies, denounces Mr. Harari.
“The will is not there. Politicians do not talk about it in the election campaign, when the danger is immediate. AI is already at work, making decisions about college admissions, about loans from banks. »
AI has the potential to help workers do some of their work, but the economic impact is still uncertain, Yoshua Bengio said.
“If AI allows a programmer to program twice as fast, are we going to need half as many programmers? Or are we going to need the same number of programmers, but they’re going to be twice as productive? It’s hard to say. »
Yuval Noah Harari notes that the transformative power of AI society is even greater than was the industrial revolution in the 19e century.
“The last great revolution was the industrial revolution. It took several generations to build the company. We’ve had a lot of bad experiences, like the Holocaust, in building functional industrialized societies. It didn’t destroy us. With the AI revolution, we cannot afford to experiment and fail, because we will not survive failures, they can destroy us. »
In the short term, he notes, many of humanity’s problems can be augmented by AI. “We don’t want to be alarmist, but the transition will be difficult. Hitler took power with 25% unemployment for only 3 years. »
Faced with the immensity of the task, Mr. Bengio notes that humans must do their best.
“It is our moral imperative. We can only try, as with climate change. There are threats, there are benefits. I am neither optimistic nor pessimistic, but we have a duty to act. »