A five-stage visual walkthrough of how an AI model gets made — data, pre-training, post-training, evaluation, and deployment — and what each stage reveals about how AI fundamentally works. No technical background required.
Data collection sets the ceiling on what a model can do. Pre-training is the big, expensive run where the model learns to predict the next piece of text. Post-training turns that base model into a helpful assistant. Evaluation and safety testing measure capability and risk. Deployment is where you finally meet the model — locked until the lab ships an update, yet sampling a different answer each time you ask.