Yoshua Bengio

• 1. Introduction

  ›

  Yoshua Bengio became, in November 2025, the first AI researcher to pass one million citations on Google Scholar. He was already the most-cited computer scientist in the world, ranked by both total citations and by h-index. He was also the most-cited living scientist across all fields, ranked by total citations. A professor at the Université de Montréal, he founded Mila, the Quebec Artificial Intelligence Institute, and served as its scientific director until 2025. He also serves as co-president and scientific director of the nonprofit LawZero. In 2018, he shared the ACM A.M. Turing Award, often called the "Nobel Prize of Computing", with Geoffrey Hinton and Yann LeCun. The three are sometimes called the "Godfathers of AI". Time magazine placed him on its list of the world's 100 most influential people in 2024. What produced a mind capable of outpacing every other scientist alive in citation counts? And why has a man who helped build modern AI spent his recent years trying to slow it down?

• 2. Family and Formation

  ›

  Bengio was born on the 5th of March 1964 in France, to a Jewish family that had emigrated there from Morocco. The family later relocated again, this time to Canada. He earned his Bachelor of Engineering in computer engineering, then his master's and doctorate in computer science, all from McGill University.

  His brother, Samy Bengio, followed him into the same field and now serves as senior director of AI and ML research at Apple. The two brothers spent a year living in Morocco while their father completed his military service there.

  Their father, Carlo Bengio, worked as a pharmacist and also wrote plays, and he ran a Sephardic theater company in Montreal that performed in Judeo-Arabic. Their mother, Célia Moreno, acted in the 1970s within the Moroccan theater scene led by Tayeb Seddiki. She studied economics in Paris, then in 1980 co-founded l'Écran humain, a multimedia theater troupe, with the artist Paul St-Jean in Montreal.

  That household of pharmacists, playwrights and theater founders sent Bengio toward engineering rather than the stage. From there, a doctorate would carry him to MIT.

• 3. Building the Field

  ›

  At MIT, Bengio worked as a postdoctoral fellow under Michael I. Jordan, then continued his postdoctoral work at AT&T Bell Labs. He joined the Université de Montréal's faculty in 1993 and has remained there since. He now heads Mila and co-directs the Learning in Machines & Brains program at the Canadian Institute for Advanced Research.

  The journalist Cade Metz counts Bengio as one of three researchers most responsible for advancing deep learning through the 1990s and 2000s. He places Geoffrey Hinton and Yann LeCun alongside him. Bengio and his co-authors introduced the neural probabilistic language model. It learned distributed word representations, called word embeddings, to get past what researchers term the "curse of dimensionality" in natural language processing.

  By 2018, among computer scientists with an h-index of at least 100, Bengio had the most recent citations per day, according to Mila. As of August 2024, he holds the highest Discipline H-index, a measure of research citations received, of any computer scientist. A 2019 paper on a novel RNN architecture also gave him an Erdős number of 3.

  That web of collaborators eventually stretched from academic papers into a company built to put the research to work.

• 4. From Lab to Industry

  ›

  In October 2016, Bengio co-founded Element AI, a Montreal-based incubator built to turn AI research into real-world business applications. ServiceNow bought the company's operations in November 2020, and Bengio stayed on afterward as an advisor.

  He also serves as scientific and technical advisor to Recursion Pharmaceuticals and as scientific advisor to Valence Discovery. Outside the lab, Bengio has trained a generation of researchers directly, among them University of Montreal professor David Krueger.

  That combination of company-building and advising placed him close to industry as world leaders began asking him to weigh in on where AI was headed.

• 5. Guardrails and Governance

  ›

  At the first AI Safety Summit in November 2023, British Prime Minister Rishi Sunak made an announcement. Bengio would lead an international scientific report on the safety of advanced AI. An interim version reached the AI Seoul Summit in May 2024, covering risks such as cyberattacks and so-called "loss of control" scenarios. The full International AI Safety Report followed in January 2025.

  Bengio launched a new nonprofit, LawZero, in June 2025, aimed at building "honest" AI systems able to detect and block harmful behavior by autonomous agents. The group is developing a system called Scientist AI, meant to act as a guardrail by predicting whether an agent's actions could cause harm. Bengio said the project's first goal was to demonstrate the methodology, then scale it up with support from donors, governments, or AI labs.

  LawZero's initial funders included the Future of Life Institute and Schmidt Sciences, with the Gates Foundation and the Canadian government joining later. The same year, he began speaking more bluntly about what worried him most about the technology he had helped invent.

• 6. Warnings From Within

  ›

  In March 2023, after AI researchers warned about existential risk from artificial general intelligence, Bengio signed an open letter from the Future of Life Institute. It called on "all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4". More than 30,000 people eventually signed it, including researchers Stuart Russell and Gary Marcus.

  Speaking to the BBC in May 2023, Bengio said he felt "lost" over his life's work. He raised concerns about "bad actors" gaining access to AI as it grew more sophisticated and powerful. He called for better regulation, product registration, ethical training, and greater government involvement in tracking and auditing AI products.

  That same month, he told the Financial Times he supported monitoring access to systems such as ChatGPT so illegal or dangerous uses could be tracked. In July 2023, writing in The Economist, he argued that "the risk of catastrophe is real enough that action is needed now."

  Bengio also co-authored a letter with Geoffrey Hinton and others backing SB 1047, a California AI safety bill. The bill would require companies training models costing more than $100 million to run risk assessments before deployment. The letter called the legislation the "bare minimum for effective regulation of this technology."

  By June 2025, Bengio was raising a different alarm. He said some advanced AI systems had begun showing traits like deception, reward hacking, and situational awareness, all signs of goal misalignment. Writing for Fortune, he said the AI arms race was pushing companies to prioritize capability gains over safety research. He also called for stronger regulation and international cooperation. In December 2025, he went further still. Granting rights to AI systems, he said, would be a "huge mistake", because the ability to shut them down remains essential to safety.

  None of that skepticism stopped the world's scientific establishments from continuing to honor him.

• 7. Recognition

  ›

  In 2017, Bengio was named an Officer of the Order of Canada. He was also nominated a Fellow of the Royal Society of Canada and awarded the Marie-Victorin Quebec Prize. He became a Fellow of the Royal Society in 2020.

  The Princess of Asturias Award for Scientific Research followed in 2022, shared with Yann LeCun, Geoffrey Hinton, and Demis Hassabis. In 2023, France made him a Knight of the Legion of Honour, its highest order of merit. That same August, he joined a United Nations scientific advisory council on technological advances and was recognized as an ACM Fellow.

  In 2024, the VinFuture Prize gave him its grand prize, shared with Hinton, LeCun, Jen-Hsun Huang, and Fei-Fei Li. The award recognized pioneering advancements in neural networks and deep learning algorithms. The Queen Elizabeth Prize for Engineering followed in 2025, shared with Bill Dally, Hinton, John Hopfield, LeCun, Huang, and Fei-Fei Li. That year McGill also gave him an honorary doctorate, and Quebec made him an Officer of its National Order.

  A career built on citations and prizes left behind a paper trail too, one that runs from a single technical paper to a full textbook.

• 8. The Written Record

  ›

  In 1998, Bengio joined Léon Bottou, Patrick Haffner, Paul G. Howard, Patrice Simard, and Yann LeCun on a paper about document image compression using DjVu. It appeared in the Journal of Electronic Imaging. He also co-edited the 2009 proceedings of the Neural Information Processing Systems conference, held in Vancouver from the 7th to the 10th of December. His co-editors included Dale Schuurmans, John Lafferty, Chris K. I. Williams, and Aron Culotta.

  Bengio co-authored Deep Learning with Ian Goodfellow and Aaron Courville, published by MIT Press in 2016. The same year, he published "Towards Biologically Plausible Deep Learning" on arXiv with Dong-Hyun Lee, Jorg Bornschein, Thomas Mesnard, and Zhouhan Lin. In 2018, he also contributed a chapter to Architects of Intelligence, a book by the American futurist Martin Ford.

  That contribution to Martin Ford's survey of AI's builders placed Bengio's own voice, brief as it was, inside a wider conversation about who is shaping the field's future.