Talking to AI — Chat Engines
Learn how a request becomes a useful brief. A model needs the task, context, audience, constraints and desired format; a confident answer can still be wrong.
Why this matters
Customer enquiries, study explanations, reports and clear instructions. A useful learning task gives you a reason to use the tool and a way to judge the result. Instead of collecting generated answers, learn to identify what the task requires and what still needs checking.
The worked task
Your practice brief
A learner needs a customer reply, but the first prompt only says “write a message.”
Use fictional or approved information. Before starting, say what a useful output would look like and which claims or actions need review.
- Write the vague prompt and record what information the answer assumes.
- Add the product, delivery terms, audience, tone and unanswered questions.
- Compare two outputs and identify unsupported promises.
- Revise the message yourself and explain why the second prompt is better.
What you can show
Two prompt versions, an edited reply and a checklist of facts to confirm.
Keep the original attempt and the improved version when appropriate. Explain which parts AI helped with, what you changed yourself and why the final output is more useful.
Review before calling it complete
- Does the result answer the actual brief?
- Which facts, calculations or behaviours did you verify?
- What is still uncertain or incomplete?
- Can you explain the important decisions without relying on the generated answer?
Connect the next skill
Compare tools on the same task without sharing private information.
Practice presenting the result or handling the conversation around it. A well-made output and a clear explanation are different skills; both deserve an attempt and feedback.
Explore the learning platform: Foundation curriculum. Examples here explain possible learning applications; they are not promises of individual results.
Explore the demo before choosing your next step.
Find an answer you can trace to a source
An AI answer sounds convincing, but the source may be missing, outdated or unrelated.
Research + verificationFollow the example →
From repeated problem to a tested prototype
A team repeatedly answers the same service questions and wants an AI assistant.
Research → design → build → testFollow the example →Questions to connect this topic with real situations
These related, illustrative practice guides can help you rehearse the conversation or task around your learning. Choose a suitable level; they do not replace the actual lesson brief or specialist review.
- I have never used AI. What should I try first?
- How can I use AI research without believing invented references?
All topic questions and scenarios · Choose a practice format