Generative AI can be a great helper at the university – writing short texts, summarising documents, translating SK/EN materials, preparing slide outlines, or drafting feedback. But the quality of the result depends heavily on how you ask. When a prompt is unclear, the model has to guess. Guessing often leads to long, generic outputs, missing facts, or even invented details. That is why learning a few simple communication habits is often more useful than learning “advanced AI features”.
Bad habits that usually produce weak results
The most common mistake is being too vague. Prompts like “Write something about GenAI”, “Make this better”, or “Create an announcement” do not tell the model what you really need. Another bad habit is forgetting the basics: the audience, purpose, and required facts. For example, if you ask for an event post but do not give date, time, place, and registration info, the model may fill gaps on its own. A third habit is not setting limits, which often causes the output to be too long, too formal, or full of unnecessary jargon.
Examples of weak prompts (and what goes wrong):
- “Write a report from the meeting.” → too generic, missing structure, unclear focus
- “Translate this and improve it.” → tone may change, meaning may drift
- “Prepare feedback for students.” → feedback may be vague or unrealistic without criteria
- “Make a presentation outline.” → might not match your audience or decision goal
Good habits that make outputs reliable and easy to use
A good prompt is usually simple, but it includes the right anchors. Start by stating what the task is and who it is for. Add what you want the reader to do (inform, invite, decide, learn). Then give a few must-have points and boundaries: length, tone, language level, and what not to do (for example: “do not invent facts”, “if something is unknown, ask or mark it as missing”). This reduces hallucinations and saves time because you fix less later.
Examples of stronger prompts (simple but clear):
- “Write a short website post (120–160 words) inviting university staff to a GenAI workshop. Use friendly, plain English. Include date/time I provide. Do not invent the room; use ‘room will be confirmed’.”
- “Summarise this text for beginners in 6 bullet points. Keep key terms. Add a one-sentence warning about checking facts.”
- “Create a slide outline for management: goal = decide whether to run a GenAI training series. Provide 8 slides max, each with 3 bullets. No marketing tone.”
- “Rewrite this paragraph in Slovak: keep the meaning, shorten sentences, remove jargon, and keep a neutral academic tone.”
A practical rule: “Context + task + limits”
If you remember only one thing, remember this: give the model context, a clear task, and a few limits. Context prevents guessing. The task tells it what to produce. Limits keep the output in the right shape. When you do this consistently, GenAI becomes much more predictable and useful – especially for everyday academic work like announcements, summaries, emails, reports, and teaching materials.
Why frameworks can help (and what comes next)
Even with good habits, people often forget something – audience, key facts, or format. That is why there are simple prompting frameworks that work like checklists. They help you communicate clearly every time, and they help teams stay consistent. In the next blog posts, we will introduce a few practical frameworks used in our workshops (e.g., CARE, RICCE, and a simple “plan → draft → check” routine) and show how to apply them to common university tasks.