Bengio — who in 2018 shared with Geoffrey Hinton and Yann LeCun the Turing Award for their work on deep learning — says he is worried about the technology leapfrogging human intelligence and capabilities in the not-too-distant future.

Speaking at his home, the professor warned that AI developments are moving at breakneck speed and risked “creating a new species capable of making decisions that harm or even endanger humans.”

OpenAI’s recent dismissal and then rehiring a few days later of chief executive Sam Altman — who has been accused of downplaying risks in his push to advance its ChatGPT bot — illustrates some of the turmoil in the startup sector and fierce competition in the race to commercialise generative AI.

For some time, Bengio has been warning about companies moving too fast without guardrails, “potentially at the public’s expense.” It is essential, he said, to have “rules that’ll be followed by all companies.” At a world-first AI summit in Britain in early November, Bengio was tasked with leading a team producing an inaugural report on AI safety. The aim is to set priorities to inform future work on the security of the cutting-edge technology.

The renowned AI academic has brought together a “critical mass of AI researchers” (1,000+) thro­ugh his Mila research institute, located in a former working-class neighbourhood of Montreal. His neighbors include AI research facilities of American tech giants Micro­soft, Meta, IBM and Google.

“This concentration of experts in artificial intelligence, which is greater than anywhere else in the world,” is what attracted Google, says Hugo Larochelle, the hoodie-wearing scientific director of the Silicon Valley giant’s AI subsidiary Deepmind.

Published in Dawn, December 3rd, 2023