Follow a simple text or image request through the chain: prompt, model, output, and response. Name what changes when you rewrite the request, rerun it, or ask for a different output format.

Compare apps and model names such as ChatGPT, Claude, Gemini, DALL·E, Midjourney, and Stable Diffusion. Distinguish the learned model parameters inside the system from request settings like output length, image size, and style instructions.

Build small prompts using task, context, audience, constraints, and desired format for text. For images, specify subject, scene, style, composition, and size so you can predict which parts of the request the output is responding to.

Apply the previous explanations in a guided problem.

Recognize tokens as the chunks used to measure prompts and outputs, not always whole words. Use the context window to reason about what the model can currently see, why long chats or pasted documents may be cut off, and how token limits affect cost and response length.

Treat fluent outputs as generated candidates, not guaranteed facts. Mark claims that need checking, notice confident unsupported details, and use provided context or outside verification to separate useful wording from hallucination.

Apply the previous explanations in a guided problem.

Check your understanding with a short quiz.

Review this chapter with practice based on your mistakes.