Yoshua Bengio

Of the three researchers who share credit for triggering the deep-learning revolution, Yoshua Bengio has arguably done the most to build lasting institutions around the technology rather than commercial ventures on top of it. Born in Paris in 1964 to a family that emigrated to Canada, Bengio completed his engineering degree, master's and PhD at McGill University, followed by postdoctoral work at MIT and AT&T Bell Labs, before joining the faculty at the Université de Montréal in 1993, where he has remained ever since.

Through the 1990s and 2000s, while neural networks were out of academic fashion, Bengio's lab produced foundational work on recurrent networks, probabilistic language modelling and, later, attention mechanisms — research that fed directly into the architectures behind today's large language models. He built the University of Montreal's machine-learning group into Mila, now one of the largest academic deep-learning research institutes in the world, drawing students and collaborators from across the globe and anchoring Montreal's reputation as a global AI hub.

In 2018, Bengio shared the ACM Turing Award with Geoffrey Hinton and Yann LeCun for their foundational contributions to deep learning, and he has since been the most consistently policy-engaged of the three. He co-founded Element AI, a commercial applied-AI lab, in 2016, but has increasingly focused his energy on governance: he signed prominent open letters urging caution on frontier model training, chaired the scientific panel behind the influential International AI Safety Report commissioned by the UK's AI Safety Summit process, and in 2025 was appointed to, then in early 2026 named co-chair of, the United Nations' Independent International Scientific Panel on AI.

In 2025 he founded LawZero, a non-profit research organisation dedicated to building safer, non-agentic 'scientist AI' systems designed explicitly to avoid the deceptive or power-seeking behaviours he has warned more autonomous AI agents could develop. He stepped back from Mila's scientific directorship in 2026, handing day-to-day leadership to Laurent Charlin while remaining as scientific advisor, a transition that freed him to concentrate on LawZero and his international advisory roles.

Bengio remains one of the most cited scientists alive in any field, and his consistent public position — that the technology he helped invent carries genuine catastrophic risk if deployed without independent oversight — has given him outsized influence over how governments in Canada, the EU and at the UN are approaching AI regulation. Few figures combine his depth of foundational technical authorship with his current standing as the field's most trusted safety convener.