Feasible
Adequate time, funding, expertise and access to data or participants.
MODULE 2
Idea Generation and Research Design
Generate and refine research ideas, explore potential research problems, formulate research questions and develop an initial research design with AI as a structured thinking partner.
QUESTION · DESIGN · VALIDATEBEFORE YOU START
Come prepared for Module 2
- Bring a broad topic or problem area you are already curious about. Even a rough one is enough to start.
- Have 2–3 papers ready to upload (from Module 1's shortlist) for the Gemini Notebook exercise.
- Create a free Google account if you do not already have one, so you can open Gemini Notebook during the session.
Idea, not evidence: Everything generated in this module, including candidate questions, FINER assessments and gap analysis, is a starting point to think with, not a finished research question. You confirm novelty and feasibility, not the AI.
Scan to open this module on your phone
drshahizan.github.io/bdc/module-2-idea-generation.html
MSO
Trainer's Tip, Dr. Shahizan
A good research question survives someone trying to break it. After Gemini Notebook suggests directions, I ask a colleague or my supervisor to poke holes in the FINER assessment before I commit weeks of work to one direction.
EXERCISE 2.1
Research Question & Design Prompt Builder ~10 min
Describe your topic and constraints. The builder assembles a ready-to-use prompt asking an AI chat tool to propose candidate research questions, assess them against the FINER criteria, and outline an initial research design.
WORKED EXAMPLE
See a completed prompt
Sample input: Topic: "use of generative AI among postgraduate researchers" · Field: Educational Technology · Context: Malaysian public university postgraduates · Type: Mixed-methods
Act as an experienced research supervisor in Educational Technology.
I am exploring this broad topic or interest area: use of generative AI among postgraduate researchers.
Target population or context: Malaysian public university postgraduates.
Preferred research type: Mixed-methods.
Propose 3 to 5 candidate research questions for this topic. For each question, assess it against the FINER criteria (Feasible, Interesting, Novel, Ethical, Relevant) in one line each. Then outline an initial research design (suggested methodology, data source and analysis approach) for the strongest question.
Important: do not fabricate literature, citations or findings. Clearly flag anything I should verify against the published literature before committing to a direction.
QUICK REFERENCE
Check every candidate question against FINER
Interesting
Engaging to you, your supervisor and the intended research community.
Novel
Confirms, extends or challenges existing findings rather than repeating them.
Ethical
Approvable by an institutional review board or research ethics committee.
Relevant
Matters to scientific knowledge, practice, or policy in your field.
RECOMMENDED TOOLS
Check whether your idea is genuinely novel
Before committing to a research question, use one of these to see what the literature already says, and whether the assumption behind your idea actually holds up.
01
CITATION CONTEXTScite
Shows whether citations support, contrast with or merely mention a publication, useful for testing the assumption behind your idea.
Visit Scite →02
RESEARCH LANDSCAPEResearchRabbit
Maps related papers, authors and citation connections for iterative exploration of your idea's landscape.
Visit ResearchRabbit →03
LITERATURE MAPPINGLitmaps
Creates citation maps and monitors new publications, a fast way to check whether your idea is already being explored.
Visit Litmaps →04
EVIDENCE SEARCHConsensus
Synthesises findings across papers and links every answer back to its source, a quick reality check for a candidate idea.
Visit Consensus →EXERCISE 2.2
Pressure-test your idea with Gemini Notebook ~15 min
Gemini Notebook builds answers only from documents you upload. Upload the handful of papers behind your idea, then ask it to compare them. This is a fast way to surface contradictions, gaps and unanswered questions before you commit to a research question.
Upload
Upload 3 to 5 key papers on your topic, either your own shortlist or highly-cited papers found with Scopus AI or Publish or Perish in Module 1.
Ask
Ask Gemini Notebook to compare the papers and surface consensus, contradictions and unanswered questions.
Refine
Use what surfaces to sharpen your candidate research question and confirm it addresses a genuine gap.
GAP-FINDING PROMPT
Compare your uploaded sources
Compare the uploaded sources on [TOPIC]. Identify: (1) areas of consensus, (2) contradictions or unresolved debates, (3) one specific gap that has not yet been addressed. Use only the uploaded sources, and flag anything uncertain or not supported.
DURING THIS SESSION
Try it with your own shortlist
- Upload 3 to 5 papers into a new Gemini Notebook, either your own shortlist or results from Module 1's Scopus AI or Publish or Perish exercise.
- Paste the gap-finding prompt above and review what Gemini Notebook surfaces.
- Note one contradiction or gap you did not expect, and consider whether it sharpens your candidate research question.
Verify before you rely on it: Gemini Notebook only reasons over what you upload, and it cannot tell you what the wider literature says. Cross-check any claimed gap against a proper search (see Module 1) before treating it as genuinely novel. See Gemini Notebook Prompts for more prompt patterns.
EXERCISE 2.3
Problem tree: from broad problem to candidate questions ~15 min
A problem tree breaks one broad problem into its root causes, and each root cause can become its own candidate research question. It stops you jumping straight to a vague RQ that is really three different questions stacked together.
- Write your broad problem or interest area at the top (e.g., "students struggle to use AI tools responsibly in coursework").
- List 3 to 4 distinct root causes that could each explain part of the problem.
- For each root cause, draft one candidate research question that addresses just that cause.
- Check each candidate question against FINER before you pick one to pursue further.
Example
Problem: Students struggle to use AI tools responsibly in coursework.
Cause 1, unclear institutional policy → RQ: How do postgraduate students interpret their university's AI use policy?
Cause 2, no verification habit → RQ: What verification practices, if any, do students apply to AI-generated content?
Cause 3, lack of training → RQ: Does a short AI-literacy workshop change students' verification behaviour?
Problem: Students struggle to use AI tools responsibly in coursework.
Cause 1, unclear institutional policy → RQ: How do postgraduate students interpret their university's AI use policy?
Cause 2, no verification habit → RQ: What verification practices, if any, do students apply to AI-generated content?
Cause 3, lack of training → RQ: Does a short AI-literacy workshop change students' verification behaviour?
PROBLEM TREE PROMPT
Ready-to-use prompt
Act as a research supervisor helping me break down a broad problem. My broad problem or interest area: [DESCRIBE YOUR PROBLEM] Field or context: [YOUR FIELD] Suggest 3 to 4 distinct root causes that could each explain part of this problem. For each cause, draft one candidate research question that addresses just that cause, and briefly note why it is feasible and researchable. Do not merge multiple causes into one question.
EXERCISE 2.4
Gap-spotting worksheet ~15 min
Papers often state a gap directly, usually near the end, and also leave gaps unstated that you can only spot by comparing several papers together. This exercise practises finding both.
- Take 3 to 5 papers from your Module 1 shortlist or literature review matrix.
- For each paper, search the discussion or conclusion for an explicit gap statement (phrases like "future research should" or "this study did not examine").
- Then look across all the papers together for an implicit gap, something none of them addressed, or a contradiction between two papers.
- Record both types of gap below, and mark which ones look genuinely novel versus already being addressed elsewhere.
| Source | Explicit gap statement | Implicit gap (across sources) |
|---|---|---|
| e.g., Smith 2021 | "Future work should test this in non-Western contexts." | No paper compares undergraduate vs postgraduate users |
GAP-SPOTTING PROMPT
Ready-to-use prompt
Act as a research assistant helping me find gaps across several papers. Papers (author, year, and a short summary or abstract of each): [PASTE YOUR PAPERS HERE] First, extract any explicit gap or "future research" statement from each paper, quoting or closely paraphrasing it. Second, compare the papers together and suggest one or two implicit gaps, things none of them examined, or points where they disagree. Base everything only on the summaries I provided, and say so if a paper does not state a gap explicitly.
Reveal a worked example
- Explicit gap: "Future research should test this framework in non-Western contexts" (Smith, 2021) is directly usable, quote it and cite the paper as the source of the gap.
- Implicit gap: If three papers all studied undergraduate samples and none studied postgraduates, that is a gap you have to spot yourself by comparing sources, not one any single paper states outright.
- Genuinely novel vs already addressed: Search for the gap topic directly (e.g., "AI literacy postgraduate non-Western") before assuming it is unaddressed. A gap may already have been closed by a very recent paper your search missed.
EXERCISE 2.5
Peer critique with the FINER rubric ~10 min
Your own research question is hard to critique objectively. Trading questions with a partner and scoring them against FINER surfaces weaknesses you would otherwise miss.
- Pair up with another participant and exchange your candidate research questions.
- Score your partner's question from 1 (weak) to 5 (strong) on each FINER criterion below.
- For any score of 3 or below, write one specific reason and one suggestion to strengthen it.
- Swap back, discuss the feedback, and revise your question if needed.
| FINER criterion | Score (1 to 5) | Comment / suggestion |
|---|---|---|
| Feasible | ||
| Interesting | ||
| Novel | ||
| Ethical | ||
| Relevant |
RELATED TOOLS & PROMPTS
Validate your idea before committing
CHECK NOVELTY
Scopus AI Prompts for Research Discovery
Explore existing trends and gaps to see whether your idea is genuinely novel.
Explore Scopus AI Prompts → SOURCE-GROUNDED Q&AGemini Notebook Prompts
More prompt patterns for synthesising and cross-checking your uploaded sources.
Explore Gemini Notebook Prompts → DISCOVERY & DATABASESAI Tools for Discovery
Search academic databases to confirm what has already been studied.
Explore Discovery Tools →WHY USE AI HERE
Traditional vs AI-assisted idea generation
| Task | Traditional approach | AI-assisted approach |
|---|---|---|
| Drafting candidate questions | Brainstorm alone or wait for a supervisor meeting | RQ Builder proposes 3–5 candidates in minutes, ready to discuss |
| Checking against FINER | Manually reason through each criterion for each question | The generated prompt scores every question against FINER automatically |
| Testing novelty | Search separately and judge by memory of what you have read | Scite or Consensus surface support, contradiction or consensus instantly |
| Stress-testing against sources | Re-read every paper to compare them against each other | Gemini Notebook compares uploaded sources and surfaces contradictions directly |
The trade-off: AI accelerates the first draft of an idea; you still confirm it addresses a genuine, ethical and feasible gap.
RECAP
Key takeaways and terms
Key Takeaways
- A strong research question is Feasible, Interesting, Novel, Ethical and Relevant (FINER).
- AI can propose and assess candidate questions quickly, but you decide which one to pursue.
- Check novelty with tools like Scite, Consensus or Litmaps before assuming a gap is real.
- Gemini Notebook only reasons over what you upload, and it cannot see the wider literature.
- FINER
- An acronym for Feasible, Interesting, Novel, Ethical and Relevant, used as a checklist for evaluating research questions.
- Research gap
- An unanswered question or unresolved contradiction in the existing literature.
- Mixed-methods
- A research design combining quantitative and qualitative data collection and analysis.
- Source-grounded AI
- An AI tool restricted to reasoning only over documents you explicitly provide.
SELF-CHECK
Test your understanding
5 MODULES · 1 DAY
See how Module 2 fits into the full programme.
Review timings for every module, breaks and the closing wrap-up.
View Full Schedule