FOLLOW THE EVIDENCE
Follow
the evidence.
Use AI to organize inquiry while keeping source support, uncertainty, and disciplinary judgment visible.
Build a trail from question to source to claim
An agent can suggest search terms, organize readings, and draft comparisons. These are useful contributions to a research workflow. Verification requires a separate look at the actual sources and the meaning of their findings. The fictional campus scenario is teaching material; research activities require real readings you have selected and checked.
Does the evidence support the claim?
Try three judgments before viewing the explanation. A calculation can be correct while the conclusion goes too far.
The starter kit contains 16 synthetic commute records: four each for walking, cycling, bus, and car. Mean journey times are 20 minutes for cycling and 24.5 minutes for cars. These records were authored for teaching.
Inspect the supplied CSV, then decide what each claim can establish.
Keep retrieval and support separate
Retrieval brings external material into the working context. It may miss relevant work, retrieve an unsuitable source, or provide a passage that supports a narrower claim than the draft makes. A real DOI or working link establishes a route to a source; it does not establish that the source supports the sentence beside it. Open the material and identify the relevant passage, page, or section. Check the population, setting, method, and qualifications. In the claim-to-source exercise, judge what the source allows you to say, then revise any wording that adds certainty or scope.
Use a ledger to make checking practical
Record the source's exact title, author or organization, year, stable link, and the claim you want to use. Add the supporting passage or a faithful paraphrase, its location, and relevant limits. Distinguish a source you have located from a full text you have read and a claim you have checked. If a document is inaccessible, leave that status visible. An agent can help populate and organize the ledger, but missing bibliographic details must remain unresolved rather than being completed from guesswork. A comparison matrix can then show where verified sources agree, disagree, or address different questions.
Treat a possible gap as an inquiry
A question unanswered by your current source set may reflect a genuine research opportunity, a narrow search, or vocabulary used differently in another discipline. Ask what the reviewed material does not answer and how you would test that absence. Expand search terms, inspect references and citing work, and consult disciplinary or library expertise when appropriate. Write a tentative statement that names the limits of the search. Discovering a relevant earlier study is a useful correction. A question can remain worthwhile even after its initial novelty claim changes. Preserve disagreements in a teaching brief so students can see how interpretation develops.
PUT THE IDEA TO WORK