The latest in AI, every dayAI News
Hugo Larochelle

AI Legend · Researcher · Scientific Director of Mila / Google DeepMind Montreal

Hugo Larochelle

The teacher who educated an entire AI generation

October 12, 2026

Some researchers change a field with a paper. Others change it by educating the next generation. Hugo Larochelle did both. This Canadian researcher from Montreal earned his PhD at the Université de Montréal under Yoshua Bengio, then completed a postdoc with Geoffrey Hinton at Toronto: a deep learning family tree that is hard to match. But what sets him apart is not just that academic lineage. It is that when deep learning was still territory reserved for a select few, Hugo opened a YouTube channel and started explaining it for free, to anyone with an internet connection.

His online tutorials on neural networks became one of the most recommended resources in the ML community: well-organized, clear, and accessible to newcomers. At the same time, his research tackled one of the hardest problems in machine learning: how does an AI learn from very few examples? His work on few-shot learning and meta-learning, where models learn to learn more efficiently, opened a line of research that now powers more capable models. He also contributed to denoising autoencoders and autoregressive models, tools that remain active parts of the modern ML toolkit.

After his doctorate he co-founded Whetlab, a Bayesian optimization startup acquired by Twitter in 2015, and worked as a Research Scientist at Twitter Cortex. He then moved to Google Brain and later Google DeepMind, leading the research team in Montreal. He was named to TIME 100 AI in 2023, sits on the boards of NeurIPS, ICML, and ICLR, and co-founded the Transactions on Machine Learning Research journal. More recently, Mila, the Quebec AI institute that Bengio co-built, named him its Scientific Director.

What impresses me most about Hugo Larochelle is not his CV, which is impressive. It is that with every door open to him, he chose the open one. Every time someone without a graduate school scholarship understands how a neural network works thanks to a YouTube video, there is a decision of Hugo's behind that moment. AI becomes democratic when the people with access decide to share what they know. That is the lesson I take from his story.


More figures who shaped AI in AI Legends, or back to the news.