A Quick Guide to Working with Chatbots
Liza Long
Before you begin
- Ethics & context check
- Is this use of AI appropriate and permitted in this context?
- Am I sharing any private data — mine or someone else’s?
- Do I understand that AI pattern-matches, not reasons? Review this chapter on how large language models (LLMs) work.
The process
- Define the problem
- What exactly am I trying to figure out, create, or solve?
- Who is this for? What does success look like?
- Is AI the right tool?
- Would a search, a textbook, or a person work better here?
- What are the potential limits of AI for this specific task?
- Craft your prompt
- Set the role, context, and constraints clearly
- Specify the format and length you need
- Use AI itself to help you refine the prompt if needed
- Test with at least two AI tools
- Run the same prompt in two different tools
- Note what differs — those gaps are revealing
- Iterate — treat your work with the chatbot as a dialogue
- Push back, ask for clarification, request examples
- Save key exchanges and note what changed between versions
- The gap between your first and final prompt is evidence of your thinking
- Evaluate the outputs
- Check claims against primary sources
- Look for corroborating expert consensus
- Flag anything time-sensitive or where the AI seemed overconfident
- Compare what the two tools produced — what differences do you see? Where are the outputs similar?
- Decide what to use and how
- Accept, modify, or reject each part—be active!
- Do you feel ownership over this work? (Can you stand behind it, explain it, and defend it as yours?)
After
- Reflect on process, thinking, and wellbeing
- Where did the AI surprise you and why?
- When did you have to bring your own knowledge to push back?
- What would have been lost if you’d stopped at the first response?
- Did this process (interacting with an LLM) energize you or drain you?
- Did AI amplify your thinking—or start to replace it?
- Are you noticing signs of overload, pressure, or fatigue around AI use?
