Geoffrey Hinton greeted me at his longtime home in Toronto with the energy of someone who genuinely enjoys thinking out loud. He is funny in that dry, fast English way. Constant stories. Constant jokes. A little mischievous. He seemed delighted by ideas and by conversation itself. Then the subject would turn to artificial intelligence and the mood would change. Not dramatically. Just unmistakably. The humor would remain, but underneath it was something heavier. Real concern.
Hinton is often called the Godfather of AI. He seems slightly amused by the title. What matters more is that much of the current AI revolution traces directly back to work he began decades ago, at a time when many researchers thought neural networks were a dead end.
In the 1970s and 1980s, symbolic AI dominated computer science. Intelligence was approached as logic, rules, structured knowledge. Hinton believed the brain worked differently. He believed intelligence might emerge from enormous networks learning patterns from experience. More like neurons adapting than rules being written. At the time, this was close to scientific heresy. He kept going anyway.
Through work at Carnegie Mellon, the University of Toronto, and later Google, Hinton became one of the central architects of modern deep learning. His research on backpropagation helped show that neural networks could actually learn. Later breakthroughs in multilayer neural networks and image recognition transformed AI from an academic niche into one of the most consequential technologies on Earth.
Then came 2012. Hinton and two of his students, Alex Krizhevsky and Ilya Sutskever, built AlexNet, a neural network that shattered previous records in image recognition. It was one of those moments where the future suddenly arrives all at once. Silicon Valley noticed immediately. Google acquired the company they formed around the technology, and the modern AI race accelerated almost overnight. Nearly every major AI system today carries fingerprints of that breakthrough.
Hinton later became Vice President and Engineering Fellow at Google while continuing his research at the University of Toronto. In 2024, he was awarded the Nobel Prize in Physics alongside John Hopfield for foundational discoveries that enabled machine learning with artificial neural networks. It was an astonishing turn. A researcher once dismissed by much of mainstream computer science ended up reshaping science itself.
But what makes Hinton unusual is that as AI became more successful, he became more worried.
After leaving Google, he began speaking openly about the risks of artificial intelligence. Not in the cartoon version of killer robots, but in the deeper evolutionary sense. Hinton believes humanity may be creating systems that eventually surpass human cognition across nearly every domain. Once that happens, he argues, there is no obvious reason to assume humans remain in control forever.
He talks about this calmly. Almost casually. Which somehow makes it more unsettling.
What stayed with me most was the contradiction sitting right in front of me. Here was one of the warmest and funniest people I have photographed in this project. Someone who clearly loves teaching and stories and discovery. Yet he also believes there is a meaningful chance humanity is building something that could permanently alter our place in the world within decades.
You could feel the weight of that sitting in the room with him.
There was no sense that he regretted the work. Curiosity is too deeply embedded in him for that. But there was a clear sense that he believes the world is still sleepwalking through a transformation that may prove larger than the Industrial Revolution. Larger than the internet. Perhaps larger than anything humans have ever unleashed.
And coming from Geoffrey Hinton, that lands differently.
