What's the first thing a beginner should learn about prompting?
Start with one clear goal
Write one sentence that says what you want back, such as a checklist or a short email. Models respond better to a concrete task than to a general topic. Your drafts will improve quickly once the goal is easy to state.
- Name the task in plain words
- Say who the output is for
- Specify the length or format
- Give one example if the style matters
Iterate instead of starting over
When the answer is off, tell the model what to change rather than rewriting the entire prompt. Small edits teach you which words control the results. Keep a short notes file of prompts that worked well.
Before you iterate, decide what a good answer looks like so you can judge the next draft fairly.
Build a habit of checking
Read each answer for facts, tone and missing pieces before you use it. A prompt can produce a confident answer that still contains errors. Treat the model as a fast first draft rather than a final authority.
Ask the model to explain its reasoning or list its assumptions when a result seems strange. Those notes reveal where your wording left room for interpretation.
Common mistakes
- Writing a vague topic instead of a task with a clear outcome.
- Judging a prompt by one answer rather than by several tries.
