Скачать или смотреть 3 Unknown Researchers Beat Google By 10%. Changed AI Forever. September 2012.

September 2012. ImageNet Challenge.

For 10 years, computers couldn't recognize images. The best teams were stuck at 74% accuracy. Nobody could break through.

Then three researchers from University of Toronto walked in:
Geoffrey Hinton
Alex Krizhevsky
Ilya Sutskever

Not famous. Not well-funded. But they had an idea everyone else abandoned: neural networks.

People said neural networks were too slow, too unstable, impossible to scale.

Hinton ignored them.

His student Alex built a deeper network - 8 layers (everyone else used 3-4).

Then they trained it using gaming GPUs. Graphics cards built for video games. Nobody else thought to use them for AI.

Results day:

Second place: 74% accuracy. Same as always.

First place: 84% accuracy.

A 10-point jump. In AI, that's not a win. That's a revolution.

Within months, every AI lab switched to deep learning. Google, Facebook, Microsoft - everyone abandoned their old approaches.

Today, every AI you use traces back to that afternoon:
ChatGPT
DALL-E
Claude
Self-driving cars
Face recognition

All built on what those three researchers proved in 2012.

The paper was called "ImageNet Classification with Deep Convolutional Neural Networks." It's been cited over 100,000 times. One of the most important papers in AI history.

AlexNet (named after Alex Krizhevsky) changed everything.

#AI #ImageNet #DeepLearning #AlexNet #GeoffreyHinton #MachineLearning #NeuralNetworks #TechHistory #AIHistory #ChatGPT #ComputerVision #Technology #Innovation #2012 #Breakthrough #Shorts

Day 15 of 1,000 - AI History
Follow for tomorrow's story