Key Takeaways: AI tools can polish your wording, but they can’t rescue a broken research question. Most Swiss university students struggle because they start with the wrong research questions. Here’s why your thesis consultations keep going in circles and the effective method to get unstuck.
Table of Contents
The Real Struggle with Building a Research Question That Works
A strong research question must be clear, focused, specific, researchable, relevant, and original. It should explore new angles or provide fresh insights. Whether you’re studying at ETH Zurich, Uni Bern, or a Fachhochschule, and are in the process of writing your thesis, the problem is the same: vague research questions waste time and energy.
Imagine you walk into your first thesis consultation with this question: “How does technology impact modern society?”
After the supervisor’s feedback: “How does AI affect university students?”
One month later: “What is the impact of AI on learning outcomes for students?”
Another month later: “How does AI influence exam performance among Swiss business students?”
One more month later: Burnout. Even after several supervisory meetings, there is still no clear direction.
The problem here isn’t laziness or lack of effort. Maybe you spent months refining your research question. At Delta Lektorat, we see this happening every semester: students torturing vague, overbroad, or impossible questions into slightly less vague versions, wasting months of their writing time and quality content on a broad research question. And here’s where most students go wrong: they turn to AI to fix the problem, which often makes things worse.
Why Sometimes AI Makes Bad Research Questions Worse
AI is a polisher, not a fixer. If you feed it a weak research question and ask it to “improve”, you’ll likely get more buzzwords that don’t add meaning, or a perfect grammar hiding an empty idea, and ten variations of the same dead-end. It’s like asking a calculator to fix your math when you don’t know the formula. The output looks neat, but the foundation is still broken.
Here are some classic research-question red flags we see:
The first is the Everything Question, which is always far too broad for a single thesis. You could write an entire PhD and still not cover it all. Then comes the Google Question, which has already been answered online countless times, leaving little room for original research. A third common misstep is the Buzzword Bingo Question, which might sound impressive but says nothing concrete or measurable. And finally, there’s the Impossible Mission Question, an idea far too ambitious and unrealistic for a bachelor’s or master’s thesis.
If your question falls into one of these categories, stop refining. Start over.
The Effective Method
Instead of asking AI to think, use this process:
Write down, honestly:
- What you really want to know (not what sounds academic)
- Why do you personally care about it
- What you can realistically study with your time and skills
Example: Instead of “I want to study AI in education,” it might be: “I’m curious why some students use AI tools like ChatGPT to improve their essays while others avoid it completely.”
Make it painfully specific.
- Generic: “How does ChatGPT affect student learning?”
- Specific: “How do students use ChatGPT when writing essays, and what impact does this have on their revision strategies?”
Now you’ve got a clear who, what, where, and when.
Test it with the “So what?” rule.
Imagine your most skeptical peer asking: “So what? Why should anyone care?” Then ask yourself:
- Can I answer this with my current skills and access?
- Do I have realistic data sources (student surveys/insights, essay drafts, supervisor feedback/insights on the topic)?
- Can I design a research method that I can finish within my thesis timeline?
If the answer is no, narrow it further.
Use proven thesis-friendly formats.
- Comparison Question: “How do students in education classes and computer science classes differ in the way they use ChatGPT when writing essays?” This lets you see patterns across different student groups.
- Change/Impact Question: “How has the way students revise their essays changed since the introduction of AI-detection software at our university?” This helps you focus on observable behaviour rather than vague quality.
- Contradiction Question: “Why do students feel more confident using ChatGPT for their essays, even though their teachers often give them lower grades?” This helps you look at the surprising gap between feelings and outcomes.
These are narrower, answerable, and help you draft a sharp research question that is also easily understood by your supervisor and peers.
Finally, use AI to scrutinise and strengthen your research questions.
Example prompt: “Evaluate my research question: ‘[insert your research question]’. What potential weaknesses or blind spots do you see, and how could it be improved for clarity, relevance, and answerability?”
AI can’t invent a solid research question for you, but it can sharpen your thinking once you’ve drafted one. You can use it to check debates, stress-test assumptions, and reveal blind spots. Think of AI as your sparring partner, not your question generator.
Conclusion
See AI as your sparring partner, not as your question generator. AI can be incredibly helpful when writing – if you use it consciously. It provides food for thought, helps with structuring, and reveals blind spots. But good academic writing emerges where humans and machines work together – with critical thinking, curiosity, and your own voice.
Sometimes, you’re simply too close to your own topic to see what’s wrong. That’s where expert help comes in. At Delta Lektorat, we don’t just polish wording. We help Swiss students build research questions that are answerable, relevant, and defensible. Because once your question is solid, the rest of your thesis doesn’t collapse under weak foundations.
Ready to stop torturing a broken research question? Upload your thesis here or contact us directly at info@deltalektorat.ch and start building your thesis on a solid foundation.
Frequently Asked Questions (FAQs)
Disclosure: This article was prepared by human contributors. Generative AI tools were used to support brainstorming, language refinement, and structural editing. All final decisions regarding content, recommendations, and academic insights reflect human judgment and expertise.
Referenzen
Qualtrics. (2025, April 5). How to write qualitative research questions (with examples). Qualtrics XM. https://www.qualtrics.com/experience-management/research/qualitative-research-question/
Litmaps. (2025, April 8). How to write a research question. Litmaps.
https://www.litmaps.com/articles/write-a-research-question
University of Alaska Fairbanks Center for Teaching and Learning. (2025, March 18). The art of crafting research question: Aligning the problem statement, goal and objectives, and research questions. UAF CTL. https://ctl.uaf.edu/2025/03/18/the-art-of-crafting-research-question-aligning-the-problem-statement-goal-and-objectives-and-research-questions/
Dea Elmasllari
Share:
