Gaming quietly built modern AI

When AI finally runs the world, you’ll know where it started. With video games. 

GPUs exist because gamers wanted more frames, better lighting, and real-time physics. 

Nvidia built parallel compute for Quake and Half-Life long before anyone cared about neural networks. 
CUDA came from graphics. Deep learning just took advantage of it later. 
Early reinforcement learning was trained in games. 
Atari environments, reward loops, fast feedback. 
Cheap, scalable, unforgiving. Perfect for learning systems. 
Demis Hassabis began as a game designer. 
DeepMind trained agents to play games before moving on to protein folding. 
Even today, AI rides on gaming demand. Gamers push GPU upgrades. Labs follow. 

Games weren’t just entertainment. 
They were the training ground.