The Université de Montréal professor supports a bill introduced in the House of Commons in June that takes a broader, beacon-based approach to AI regulation, but many of the details remain to be worked out.
Ottawa pointed out that the C-27 bill will not take effect until 2025, a period that Bengio considers slow, as he pointed out this May 24 in a press conference on the sidelines of a C2 conference in Montreal.
It’s too slow a pace. There are simple things that could be implemented and for which two years of reflection are not needed
he explained.
“Counterfeit Humans”
The artificial intelligence expert called on the federal government to immediately start enforcing rules against certain threats, such as falsification of humans
which employs AI-driven robots that impersonate people.
Users need to know if they are talking to a machine or a human. Social media accounts need to be regulated so we know who is behind the account, and they have to be human most of the time
said Bengio, who won the Turing Award in 2019, a reward widely considered the technology industry’s Nobel Prize.
Bengio cautioned that AI systems, including those that incorporate conversational tools like ChatGPT, represent a existential risk
for the humanity.
Criticized for its vagueness by some legal scholars, the liberal proposal for an Artificial Intelligence and Data Act seeks to establish a framework for responsible AI development that seeks agility as technology rapidly evolves, while prohibiting its malicious use. That proposed law would establish an economic sanctions and oversight body.
The government intends to give the ecosystem enough time to adapt to the new legal framework before taking coercive measures.
Canada’s Ministry of Innovation, Science and Economic Development said in a document accompanying the bill.
“If today’s AI is like an amoeba, just imagine what it would be like as a Tyrannosaurus,” explained historian Yuval Noah Harari.
Photo: Reuters / Dado Ruvic
An existential risk to humanity
Speaking with historian Yuval Noah Harari at the C2 conference in Montreal, Bengio agreed with his warning that AI systems represent an existential risk to humanity
and democracy.
Harari, author of the book Sapiens: a brief history of humanity
considers that the pace and magnitude of the development of generative AI exceeds the technological advances of previous eras.
We are up against something even more powerful than the trains, radio, and electricity we invented in the industrial revolution. I believe there is a way to build good societies with AI, but it will take time.
Harari said in a video conference, warning that failed experiments will not leave room for a second chance: we will not survive
.
In March, the two speakers joined more than 1,000 artificial intelligence experts in calling for a six-month pause in training AI systems more powerful than GPT-4, the large underlying linguistic model in OpenAI-based ChatGPT. in San Francisco.
Among the people who signed that request for a pause in the development of artificial intelligence included engineers from Amazon, Google, Meta and Microsoft, as well as Apple co-founder Steve Wozniak and Rachel Bronson, president of the Bulletin of Atomic Scientists.
Challenges for democracy
If the current AI is like an amoeba, just imagine what it would be like as a tyrannosaurus. And it won’t take billions of years to get there.
illustrated Harari in his conversation with Bengio.
In a way, propaganda, misinformation and custom trolls that can influence your vote
they will pose a challenge to democracy in the next decade, Bengio said.
Safeguards need to be put in place as soon as possible
he added. We have to face the dangers, the sooner the better
.
Source: RC/PC
Adaptation: RCI / R. Valencia