Prompt like a PRO: How to Get the Best Result from AI New
AI tools have become extremely powerful and entered very many fields - from coding and design to marketing, education and writing. The fact is AI cannot read your mind (yet), so the quality of what you get from any tool heavily depends on how well you ask.
Photo by DC Studio on Freepick
By Sofia Stalskaya
Senior Lecturer at Department of Foreign Languages
HSE University in St. Petersburg
Whether you are a researcher and educator or just a curious user, mastering the skill of prompting will help you unlock more accurate results.
Why does prompting matter? Just think of AI as a student.
If you provide vague instructions, you will get a vague result.
If you are precise in guidance, you will get what you expect.
Good prompting saves time, improves the output result and reduces frustration about tasks.
AI is a Tool, not a Mind Reader.
AI does not understand the situations as we, humans, do. It generates responses based on patterns it learned during training. So, your prompt has to be more of an instruction or algorithm to achieve better results.
A great prompt usually includes
- clarity
- context
- format
- tone and style
- constraints
Let’s have a closer look at each component.
Clarity. The point is not to leave any room for misinterpretation, so the prompt executes exactly what you ask for. Irrelevant requests or the wrong language might lead to incomplete results. So, think of clarity like writing a detailed protocol, every step should be explicit.
❌ "Summarize the paper [paste text/PDF]"
✅ "Extract key findings, methodology, and limitations from this research paper [paste text/PDF]. Format as a bullet-point summary."
Using active verbs like “summarize” is good but not enough. You have to be specific about the criteria of the best outcome, that’s why being precise about key findings, methodology, and limitations seems a better option.
Context. Providing the background information helps AI understand the specific situation, field, and requirements. Without the context the algorithm might be too general and even shallow.
❌ "Write a motivation letter for a conference"
✅ "I am a PhD candidate in Linguistics, researching into conceptual metaphor theory and its application to bilingual cognition. I am applying to attend the International Conference on Cognitive Linguistics (ICCL) taking place in [Location] in [Month Year]. Write a compelling motivation letter (300-400 words) that will convince the conference organizers to accept my application for attendance."
The more details you provide - background, field of study, papers names, audience, and key points - the better the output.
Format. Specifying the format helps AI structure the output in a way that is immediately useful, whether it's a list, a paragraph, a table, or an outline. Without format guidance, the result may be correct but not convenient.
❌ "Give me tips for writing a literature review."
✅ "Provide a step-by-step guide (numbered list) on how to structure a literature review for an undergraduate’s Project Proposal. Include sections on thematic organization, source synthesis, and gap identification."
Being focused on format, the prompt still presents clarity and includes the background. So, it is never only one feature taken into consideration, it is always a combination of several.
Tone and Style. AI models are trained on diverse content, so without guidance, they may default to a tone that does not quite match your purpose. It might be too casual or too academic. Specifying the tone and the style ensures the output aligns with your voice and audience expectations. The prompt should reflect the level of formality as well as terminology appropriate for your purpose. For example, if you are working on some activity for your students, the output has to be different from the one you expect for your own academic purposes.
❌ "Draft an email to a journal editor."
✅ "Draft a 150-word polite and professional inquiry email to a journal editor regarding the current status of my submitted manuscript. Do not ask for a specific timeline, but express eagerness to receive an update. Include a subject line."
Constraints. Setting boundaries is essential for precision, it helps to keep AI on track and avoid irrelevant or overly broad responses. If you look at the example above, you might notice a specific instruction concerning the number of words needed. These limitations can also include not only the word count, but any other specific exclusions.
Overall, AI is not a miracle worker, it is a powerful tool that responds to thoughtful input. The more effort you put into crafting your prompt, the more valuable the output becomes.
Meta-Prompt: Let AI Help You Prompt
Taking into consideration all mentioned above, I should say that there is certainly another option of how to manage your communication with a model which works perfectly well if you already understand the peculiarities of building the right prompt. Crafting the prompt by yourself is not always necessary, though understanding of how it is done is crucial for evaluating the result. Once you know how to write a good prompt, there's one more way to guide the model.
To save time, you can always ask a model itself to write a prompt for you. This is called a “meta-prompt” and understood as a prompt that you use to generate other, better prompts. Instead of asking the AI to perform a task directly, you ask it to act as an expert prompt engineer and create an optimal instruction for your specific need.
✅ Act as an expert prompt engineer. I need to generate a detailed lesson plan on _______ . Write an optimal prompt to get the best possible output.
My final recommendation would be to think of prompting not as a technical skill, but as an extension of the clarity of thought we value so much in an academic environment. A well-structured prompt forces us to define our own goals and audience more precisely. In this way, learning to prompt effectively doesn't just make us better at using AI, it helps us become a better communicator overall.
Tools Used:
-
Merlin.Ai. (2025). Artificial Intelligence Tool for Writing and Editing Texts. Retrieved September 15 , 2025, from https://www.merlin.ai
-
Qwen.Chat. (2025). Artificial Intelligence Tool for Learning and Communication. Retrieved September 15, 2025, from https://chat.qwenlm.ai/