Скачать или смотреть How YouTube Cat Videos Accidentally Created Modern AI

Millions of people uploaded random videos to YouTube for years — cats, memes, funny clips, and everyday moments. Most of it looked like useless internet noise. But in 2012, that “noise” became the fuel for one of the biggest breakthroughs in artificial intelligence. This video tells the real story of how Google Brain and Andrew Ng trained a neural network on massive unlabeled YouTube data… and the machine learned something unexpected first: it learned to recognize cat faces. That single discovery proved a new idea: AI doesn’t always need humans to label everything. With enough raw data, patterns can emerge on their own. But there was a problem. This kind of training required insane computing power, something only tech giants could afford. Then came the next revolution: GPUs. Originally made for video games, GPUs turned deep learning from impossible to scalable, making modern AI systems faster, cheaper, and accessible to the world. From the “cat neuron” experiment to Nvidia’s rise and today’s large models like ChatGPT, this episode explains how data, hardware, and timing combined to change history — and why one of the biggest AI moments started with something as simple as cat videos.