Use the suggestions to choose a direction for preliminary reading. A plausible question is not proof that the topic is new, that suitable sources exist, or that a study can answer it. Follow your course or institution's AI policy before using the tool for this work.
Bring the constraints that change the question
The following original fictional assignment is a two-week, 1,500-word class paper about public-library reservation interfaces. The student can inspect published help pages and read library and interface research. They cannot recruit participants, access private usage logs, or run a controlled experiment for this assignment.
Those constraints make some questions more feasible than others:
| Proposed question | Evidence it would require | Fit for this assignment |
|---|---|---|
| Does a redesigned reservation screen increase older adults' successful reservations? | A defined redesign, an appropriate study population, a defensible comparison and a measured outcome | Does not fit the stated document-based assignment |
| How do selected public libraries' help pages explain the steps and prerequisites for reserving an item? | A stated set of current public help pages and a consistent way to compare the instructions | A feasible candidate if enough suitable pages can be found |
| What usability barriers do published studies report for older adults using public-library reservation interfaces? | Relevant research that actually addresses this population and task | Worth a preliminary search; source availability remains unknown |
These alternatives ask different kinds of questions. The second examines documents, while the third reviews reported research. Neither permits a conclusion about how all older adults use library systems.
UC Davis Library's research-question guidance connects scope with the terms and themes used in existing research. Use initial reading to revise your wording. “Reservation” may not be the term your relevant papers use; inspect the actual texts before adopting suggested synonyms.
Ask for alternatives with their evidence requirements
OpenAI's prompting documentation recommends giving the relevant context and describing the result you need. For question development, useful context includes what you can investigate, not only what interests you.
Help me develop research questions, not an answer or a thesis claim.
Topic and what interests me: [ENTER]
Assignment type, length, deadline and AI-use rules: [PASTE]
Sources, methods and permissions actually available: [ENTER]
Relevant preliminary reading with source references: [PASTE OR NONE]
Suggest three substantially different, neutral questions. For each,
explain its scope, what evidence would answer it, and which required
evidence I do not yet have. Distinguish a document comparison, a
literature-based question and a question needing new empirical data
where relevant. Do not invent papers, novelty claims, populations,
data access or ethics approval. Do not assume a causal effect.
Propose preliminary search concepts, labeled suggestions to check.
Ask which tradeoff I prefer before expanding one into a working plan.
Reject a question that smuggles its conclusion into the wording. “Why are instructions inaccessible?” presumes the accessibility finding. “Which steps and prerequisites do these pages state?” leaves the finding open, though it answers a narrower question.
Test the candidate with preliminary searches
Search the library catalog, relevant databases, or the actual document collection available to you. Record search terms, places searched, and a few sources you have opened. Ask whether they address the same population, task, and outcome as the proposed question.
If the search finds only general library-app studies, you may need to broaden the interface task or change the population boundary. That is a research choice to discuss with your instructor, rather than a reason to describe general studies as reservation-specific evidence.
For the document-comparison candidate, identify the selection rule before choosing attractive examples. “The three pages easiest to find” may be acceptable for a limited class exercise if stated honestly; it does not represent every public library. Confirm whether the assignment expects a different sampling approach.
Leave the question open to what you find
Save a working question, its scope, the sources you located, and the remaining feasibility issue. The literature-review matrix can help relate sources after you have read them. If you later ask ChatGPT to investigate the topic, the research-report guide covers setting those research boundaries.
For now, take the candidate question and two or three actual source examples to your instructor. Agree on what the paper can reasonably answer before turning an interesting topic into a claim.