Applied AI Lessons

Experiment 2 of 6

Run a prompt-and-evidence audit

Investigate a small question with two prompt versions and a source trail you can defend.

Choose a focused question that can be investigated with two or three sources. Write a deliberately simple initial prompt.
Before generating, define the audience, goal, acceptable evidence, and format.
Revise using Concise, Logical, and Explicit. Compare the results using the same criteria.
Open three sources. Record bibliographic details and a supporting passage for each key claim.
Adapt the prompt to address one weakness and reflect on what changed.

Starting prompt

Help me explore [focused question] for [audience]. Produce a short outline with competing explanations and search terms. Separate claims from hypotheses. If you provide sources, give enough detail for me to find them and state whether you actually accessed them. Do not fill evidence gaps with invented references.

Reflect

Which improvement came from better direction, and which required better evidence?

Don’t include private student, research, or workplace information. No paid tool needed: Use library search directly and compare two librarian-provided sample AI answers. Analyze the same prompting and evidence decisions.

Finish this experiment

Write a short reflection: what you made, what you changed, and what you learned. At least 80 characters.

Tools to try: ChatGPT, NotebookLM, Perplexity, Elicit

Run a prompt-and-evidence audit

Starting prompt

Reflect

Finish this experiment