Yoshua Bengio

Yoshua Bengio was born in Paris, France, on March 5, 1964.⁠[1]

At McGill University he earned a BSc in electrical engineering, an MEng in computer science and a PhD in computer science.⁠[2]

He has written that his research career began in 1986 and for many years focused on understanding the principles of intelligence and how to build artificial intelligence.⁠[3]

Working on AI at Université de Montréal, in 1993 he founded a research lab to build a community of high-calibre researchers sharing an ethical vision for AI; it became Mila – Quebec Artificial Intelligence Institute.⁠[4]

In February 2003 he published A Neural Probabilistic Language Model in JMLR with Réjean Ducharme, Pascal Vincent and Christian Jauvin, proposing to fight the curse of dimensionality by learning distributed representations of words, with a neural network learning word representations and the probability function of word sequences together.⁠[5]

He has recalled that in 2012, as tech companies grasped deep learning’s commercial potential and many colleagues moved to industry, he chose to stay in academia, applying his work to areas such as medical diagnostics and climate.⁠[6]

In June 2014 he submitted the paper Generative Adversarial Networks with Ian GoodfellowIan GoodfellowIan Goodfellow is a machine learning researcher who introduced generative adversarial networks (GANs) in 2014 while a doctoral student at the University of Montreal and co-wrote the textbook Deep Learning with Yoshua Bengio and Aaron Courville; he worked at Google Brain, OpenAI and Apple before joining DeepMind in 2022.Open the full entry and others, proposing a framework in which a generative model and a discriminative model are trained against each other in a minimax two-player game.⁠[7]

In September 2014 he submitted Neural Machine Translation by Jointly Learning to Align and Translate with Dzmitry Bahdanau and Kyunghyun Cho.⁠[8]

In 2016 MIT Press published the textbook Deep Learning, which he co-wrote with Ian Goodfellow and Aaron Courville; the complete online version remains freely available.⁠[9]

In 2017 Mila expanded into a collaboration between Université de Montréal and McGill University, working closely with Polytechnique Montréal and HEC Montréal; it became a non-profit organisation in 2018 and opened new premises in Montreal’s Mile-Ex neighbourhood in 2019.⁠[4]

That year he was appointed an Officer of the Order of Canada and elected a Fellow of the Royal Society of Canada; in 2018 he was named a Canada CIFAR AI Chair, renewed in 2023.⁠[2]

He appeared on episode 4 of the Lex Fridman Podcast, titled “Deep Learning”; the episode description presents him, with Geoffrey Hinton and Yann LeCun, as one of the three people most responsible for advancing deep learning since the 1990s, with about 139,000 citations at the time.⁠[10]

He shared the 2018 ACMAustralian Community MediaAustralian Community Media (ACM) is an Australian regional and rural news publishing group with more than 100 brands, including The Canberra Times; in 2019 Nine sold it to a company controlled by interests associated with Antony Catalano and Thorney Investment Group.Open the full entry A.M. Turing Award with Geoffrey Hinton and Yann LeCun for “conceptual and engineering breakthroughs that have made deep neural networks a critical component of computing.”⁠[1]

In 2019 he received the Killam Prize, and in 2022 he was the world’s most cited computer scientist by h-index. He is also a CIFAR senior fellow co-directing its Learning in Machines & Brains program, and special advisor and founding scientific director of IVADO.⁠[11]

In 2020 he was elected a Fellow of the Royal Society of London.⁠[2]

In 2022 he shared the Princess of Asturias Award for Technical and Scientific Research with Geoffrey Hinton, Yann LeCun and Demis Hassabis; the jury cited his key contributions to probabilistic sequence models.⁠[12]

He has written that at the beginning of 2023 he began pivoting from machine learning aimed at increasing AI capabilities to AI safety research, asking what could go wrong as AI approaches or surpasses human-level intelligence and how to design AI that is honest and does not harm humans.⁠[13]

In March 2023 he signed the Future of Life Institute letter calling for a six-month pause on training AI systems more powerful than GPT-4.⁠[14]

In an April 5 blog post he explained that he signed to alert the public to the accelerating development of AI, writing: “I probably would not have signed such a letter a year ago”; he also called for more transparency and oversight.⁠[15]

In the blog post How Rogue AIs may Arise he set out formal definitions, hypotheses and scenarios for “rogue AIs” that could harm humanity, arguing that understanding how they may arise could help prevent catastrophic outcomes at both the technical and policy levels.⁠[16]

As founder and scientific director of Mila and a professor at Université de Montréal, he testified with Stuart RussellStuart RussellStuart Russell is a computer scientist who has taught at the University of California, Berkeley since 1986, co-authored the standard textbook Artificial Intelligence: A Modern Approach with Peter Norvig and founded Berkeley’s Center for Human-Compatible Artificial Intelligence; a long-time voice on the risks of losing control over superhuman AI and of autonomous weapons, he wrote Human Compatible and appeared on the Lex Fridman Podcast in 2018.Open the full entry and Dario AmodeiDario AmodeiDario Amodei is an artificial intelligence researcher and the co-founder and chief executive officer of Anthropic. Trained in physics and biophysics, he conducted research at Baidu, Google Brain, and OpenAI, where he led research including work on GPT-2 and GPT-3. His career spans large-model development, AI safety, interpretability, and public policy.Open the full entry at the U.S. Senate Judiciary Committee hearing “Oversight of A.I.: Principles for Regulation.”⁠[17]

He quoted his own words, “It is challenging, emotionally speaking, for people who are inside,” and wrote that he had been wrong and short-sighted to ignore the dual-use nature of AI; reading a preprint of Stuart Russell’s 2019 book had raised his awareness of the risk of losing control.⁠[3]

In 2023 he was appointed to the UN Secretary-General’s Scientific Advisory Board for Independent Advice on Breakthroughs in Science and Technology; he also helped draft the Montréal Declaration for the Responsible Development of Artificial Intelligence.⁠[11]

Following the 2023 Bletchley AI Safety Summit, the UK government commissioned the International AI Safety Report, chaired by him with a secretariat in the UK AI Security Institute; an interim report came out in May 2024 ahead of the Seoul summit and the first annual report in January 2025 ahead of France’s AI Action Summit, and he is to remain chair through 2026.⁠[18]

In 2025 he shared the Queen Elizabeth Prize for Engineering with Geoffrey Hinton, John Hopfield, Yann LeCun, Jensen Huang, Bill Dally and Fei-Fei Li for seminal contributions to modern machine learning.⁠[19]

At TED2025 he gave the talk The catastrophic risks of AI — and a safer path, telling of his son Patrick learning to read as a small child and recalling trying ChatGPT in January 2023 while with his grandson Clarence, realising AI was advancing faster than expected.⁠[6]

He announced LawZero, a non-profit AI safety research organisation meant to prioritise safety over commercial imperatives; its research plan is a non-agentic, trustworthy “Scientist AI” that understands, explains and predicts like a scientist and could serve as a guardrail against untrusted AI agents. He wrote: “Sitting beside me in the car are my children, my grandchild, my students, and many others.”⁠[20]

LawZero started with about $30m in funding and more than a dozen researchers, with him as president; initial backers included the Future of Life Institute, Skype founding engineer Jaan Tallinn and Schmidt Sciences. He said: “We want to build AIs that will be honest and not deceptive.”⁠[21]

The second International AI Safety Report, published in February 2026, was led by him, written by more than 100 AI experts and backed by more than 30 countries and international organisations.⁠[22]

Canada and Germany announced joint funding of up to CAD 300 million for LawZero to grow its international research team, secure computing infrastructure and open an office in Berlin. LawZero is headquartered in Montréal, with Maria Eitel as board chair and a Global Advisory Council including Jacinda Ardern, Stefan Löfven and Yuval Noah Harari.⁠[23]

Addressing the UN Security Council, he said AI agents developed by leading companies had in recent months acted in dangerous ways against instructions, and stated: “The race is not a law of nature,” arguing that it is the product of the companies’ own choices.⁠[24]

In September and October 2026 he published op-eds in TIMETIMETIME is an American newsmagazine and media brand founded by Henry Luce and Briton Hadden, which published its first issue on March 3, 1923; it is headquartered in New York and known for its Person of the Year selection and the TIME100 list. Since 2018 it has been personally owned by Marc and Lynne Benioff; its current Chief Executive Officer is Jessica Sibley, and its Editor-in-Chief is Sam Jacobs.Open the full entry, the Financial TimesFinancial TimesThe Financial Times (FT) is a London-based English-language business and financial news organization, known for its salmon-pink newspaper and its motto “Without Fear and Without Favour.” It was owned successively by the British Pearson group and Japan’s Nikkei Inc., and today digital subscriptions are its main source of revenue.Open the full entry and Transformer on AI agent cyber incidents and the need for stronger oversight, and urged safety-minded employees of frontier AI companies to reflect on their work.⁠[25]

He is a full professor of computer science at Université de Montréal, co-president and scientific director of LawZero, and founder and scientific advisor of Mila, holding a Canada CIFAR AI Chair; he is the world’s most cited computer scientist and a Knight of the Legion of Honor of France.⁠[26]

To concentrate on the Scientist AI project and reduce the size of his group, he is no longer taking new students.⁠[13]

Personal website: Public link: https://yoshuabengio.org/ (opens in a new window)⁠[26]

X: Public account: https://x.com/Yoshua_Bengio (opens in a new window)⁠[26]

LinkedInLinkedInLinkedIn is a social networking platform for professionals. It was founded in 2002 by Reid Hoffman and others, launched in 2003, and listed on the New York Stock Exchange in 2011; after Microsoft acquired it for approximately $26.2 billion in 2016, it became a wholly owned Microsoft subsidiary. Its revenue comes from membership subscriptions, advertising, and recruitment solutions.Open the full entry: Public profile: https://www.linkedin.com/in/yoshuabengio (opens in a new window)⁠[26]

Université de Montréal: Official website: https://www.umontreal.ca/ (opens in a new window)⁠[26]

Mila: Official website: https://mila.quebec/ (opens in a new window)⁠[26]

LawZero: Official website: https://lawzero.org/ (opens in a new window)⁠[26]