Nib asked, and Nib — askNib's tutor — drew the answer live at a whiteboard. This is the spoken transcript; enable JavaScript to watch it drawn.
How do AI language models like ChatGPT actually work? (for a curious adult who uses chatbots but has never studied computing). Cover, in this order: How does ChatGPT work?; How does it read your words?; How does it learn which words come next?; How does it know which earlier words matter?; Why do AI chatbots hallucinate?; Why does it forget what you said earlier?; How did a text predictor learn to follow instructions?; Does it understand, and how much should you trust it? — page 6: The Context Window Limit
Every prediction looks at the text so far. But there's a ceiling on how much text fits in view. That ceiling is the context window.
Everything shares one budget: your messages, its replies, hidden instructions, even the answer it's writing right now. Long replies crowd out earlier text.
When the chat outgrows the window, the oldest text slides out. The model isn't ignoring it. For the next prediction, it simply isn't there.
The app doesn't store your chat. Each turn it resends the whole conversation, and the model rereads it from scratch. No resend, no memory.
So repeat what matters, close to where you need it. In a long chat that's drifting, start fresh and paste in a short summary.
The model isn't remembering you; it's rereading a limited stretch of text each time, so what you want it to use must be inside that window.