AI Skill Report Card
Generating Research Questions
Quick Start14 / 15
Given a source text, produce 4-6 research questions, each with 3-4 categorized "angle" answers (comparative, causal, evaluative, policy). Output format:
Q: [Open-ended research question derived from text]
- [Angle label] (comparative): [One-sentence claim comparing two approaches/methods]
- [Angle label] (causal): [One-sentence claim asserting cause→effect]
- [Angle label] (evaluative): [One-sentence claim judging effectiveness/tradeoffs]
- [Angle label] (policy): [One-sentence recommendation for institutions/practitioners]
Recommendation▾
Add a third example with a different domain (non-textile) to prove generalizability across topics, since both current examples use fashion/manufacturing content
Workflow14 / 15
Progress:
- Step 1: Read source text; extract 4-8 key concepts, tensions, or transitions (e.g., "manual vs. digital", "cause X enables effect Y")
- Step 2: For each concept/tension, draft one open-ended research question (not yes/no)
- Step 3: Tag each question's implicit scope (Specific/General/Broad) and complexity (Basic/Moderate/Advanced/Expert) — favor Moderate-to-Advanced for source texts with technical depth
- Step 4: For each question, generate 3-4 short "angle" answers, one per type:
- Comparative — contrasts two methods, eras, or systems mentioned in the text
- Causal — claims a mechanism-driven cause-and-effect relationship
- Evaluative — makes a judgment call about effectiveness, tradeoffs, or limitations
- Policy — recommends an action for an institution, firm, or field
- Step 5: Give each angle a 2-5 word descriptive label (e.g., "Standardization vs. artisan nuance")
- Step 6: Review that questions cover the full text (don't cluster all questions around one paragraph)
- Step 7: Present final list, question-by-question, angles indented beneath
Recommendation▾
Include an edge case example showing a short or ambiguous source text and how to handle insufficient material for 4-6 questions
Examples17 / 20
Example 1: Input: Paragraph about CAD software optimizing textile marker layouts and reducing material waste. Output:
Q: In what ways does CAD software influence the amount of fabric waste generated during production?
- CAD and waste reduction (causal): The implementation of CAD software significantly reduces fabric waste by allowing designers to optimize pattern nesting through real-time simulation.
- Efficiency of CAD versus manual methods (comparative): Compared to manual patternmaking, CAD systems decrease scrap rates for complex designs but offer negligible gains for simple, modular pieces.
- CAD effectiveness against design-based reduction (evaluative): Reliance on CAD for fabric efficiency is less effective than adopting zero-waste pattern-cutting techniques that redesign the production process itself.
- Policy for manufacturing workflows (policy): Manufacturing firms should prioritize integrated CAD-cutting table workflows to minimize waste, as linked systems offer greater precision than standalone software.
Example 2: Input: Sentence noting that 3D visualization tools let patternmakers simulate fit and drape before cutting fabric. Output:
Q: To what extent is 3D visualization technology effective at reducing the need for physical prototypes?
- Limitations in complex tailoring (evaluative): While 3D visualization reduces early-stage iterations, it remains insufficient for complex tailoring requiring tactile evaluation of drape and structural integrity.
- Efficiency gains in mass-market production (comparative): Adopting 3D simulation as a primary tool decreases waste and costs more effectively than physical sampling cycles in mass-market apparel.
- Workflow transformation (causal): Integration of 3D virtual prototyping shifts construction from a physical-testing model to a digital-validation model, shortening lead times.
- Hybrid prototype strategies (policy): Firms should adopt a hybrid approach where 3D visualization replaces initial drafts but physical prototypes remain mandatory for final QA.
Recommendation▾
Clarify how scope/complexity tags (mentioned in Step 3) should appear in final output, since the Quick Start format and examples omit them entirely
Best Practices
- Derive every question from an explicit claim or transition in the source text — don't invent unrelated topics.
- Keep questions open-ended: prefer "In what ways...", "To what extent...", "How does..." over yes/no phrasing.
- Ensure the four angle-types per question are genuinely distinct in reasoning style, not just reworded restatements of each other.
- Keep each angle answer to one sentence — a claim, not a paragraph.
- Distribute questions across the entire source text so no single paragraph is over-represented.
- Match complexity to the source: technical/industry text warrants Advanced or Expert-level questions.
Common Pitfalls
- Don't produce simple factual/definitional questions ("What is a sloper?") — target analytical, discussion-worthy questions.
- Don't let the "policy" angle default to vague advice ("more research is needed") — tie it to a specific institutional or industry action.
- Don't repeat the same angle content reworded across comparative/causal/evaluative — each must use its distinct reasoning mode.
- Don't fabricate specifics (brand names, statistics) not implied by the source text.