Contents
Prompt Engineering
Last updated:
|
|
Prompt engineering (prompting) is…
The technique of designing effective questions or instructions, known as "prompts" to elicit clear, relevant responses from AI models (Jacobsen & Weber, 2023).
Imagine you are giving directions to an assistant to help you with a task – prompt engineering is essentially playing around with different ways to structure and phrase your questions or commands to make them more understandable for AI to perform the task at hand. Here are two examples:
Weak prompt |
Strong prompt |
“Tell me about hedging.” |
“I am learning about academic writing. Explain the concept of hedging in simple terms, including three examples of hedging language and how it can be used to express uncertainty or moderate claims." |
The clarity and accuracy of the prompts AI receive determine the quality of the output. Providing more context and details to AI will help it understand your instructions better. There are many different ways to construct effective prompts.
The CIDI framework
For example, your prompts can be structured around these elements following the CIDI framework (Ryan, 2024):
1. Context
Define the role of AI in your task. e.g. "You are an Engineering student presenting a mechanical innovation in class." This helps guide the AI's perspective and approach.
2. Instructions
Outline the specific tasks you want AI to perform. Be precise. If you need AI to help summarize an article, specify what elements should be included, such as key findings, the method used, limitations etc. E.g. “Summarize the article on the impact of global warming, including the main argument, key findings, methodology, limitations, and any recommendations made by the authors.” If you want more specific output from AI, try giving it some guiding questions.
E.g.
“Please summarize the following article on global warming. Include the following elements in your summary:
Key findings: What are the main results or conclusions regarding the effects of global warming presented in the article?
Methodology: What specific research methods (e.g., surveys, experiments, data analysis) were employed to gather data on global warming?
Limitations: What limitations of the study regarding global warming are mentioned by the authors?
Implications: What do the authors suggest for future research or practical applications related to global warming based on their findings?”
3. Details
Include any important details that frame how the output should be presented. For example, if the report should focus on a specific cultural context or historical period, make that clear. Include constraints or requirements, like "Use a formal tone and academic vocabulary".
4. Input
Give AI the input it needs to perform the task. This might include data sets, documents, previous reports, or any relevant background information. For instance, “Write your summary based on this article: Marzuki, Widiati, U., Rusdin, D., Darwin, & Indrawati, I. (2023). The impact of AI writing tools on the content and organization of students’ writing: EFL teachers’ perspective. Cogent Education, 10(2), 2236469, DOI: 10.1080/2331186X.2023.2236469”.
Let’s take a look at this example. In this scenario, a student wants AI to help him/her write a summary paragraph for an article, so he/she structures the prompt using the CIDI framework:
Context: You are a university student writing a literature review for an essay about artificial intelligence.
Instructions: Write a short summary (150 words) for this article (Marzuki et al., 2023), including these elements: A brief description of the study, methods used, sample size, key findings, critical analysis of the limitations, and implication of the results.
Details: Use a formal and academic tone. Use APA 7th edition citation format and cite all non-original ideas.
Input: Write your summary based on this article: Marzuki, Widiati, U., Rusdin, D., Darwin, & Indrawati, I. (2023). The impact of AI writing tools on the content and organization of students’ writing: EFL teachers’ perspective. Cogent Education, 10(2), 2236469, DOI: 10.1080/2331186X.2023.2236469
Instructions: Write a short summary (150 words) for this article (Marzuki et al., 2023), including these elements: A brief description of the study, methods used, sample size, key findings, critical analysis of the limitations, and implication of the results.
Details: Use a formal and academic tone. Use APA 7th edition citation format and cite all non-original ideas.
Input: Write your summary based on this article: Marzuki, Widiati, U., Rusdin, D., Darwin, & Indrawati, I. (2023). The impact of AI writing tools on the content and organization of students’ writing: EFL teachers’ perspective. Cogent Education, 10(2), 2236469, DOI: 10.1080/2331186X.2023.2236469
See AI’s response (DeepSeek R1, accessed 3 APRIL 2025)
Marzuki et al. (2023) investigated English as a Foreign Language (EFL) teachers’ perceptions of AI writing tools’ influence on students’ writing content and organization. Using a qualitative approach, the researchers conducted semi-structured interviews with 15 Indonesian EFL teachers to explore their experiences and observations. Findings revealed that while AI tools enhanced students’ writing efficiency, structural coherence, and vocabulary diversity, they also led to over-reliance, reduced critical thinking, and formulaic outputs. Teachers noted challenges in assessing originality and fostering creativity. A critical limitation is the small, context-specific sample, which restricts generalizability. Additionally, the study’s reliance on self-reported data may introduce bias, and the absence of student perspectives limits holistic understanding. The results imply that AI integration requires balanced pedagogical strategies to harness benefits while mitigating dependency. Educators should prioritize scaffolding critical thinking alongside AI use. This study underscores the need for further research on longitudinal impacts and cross-cultural comparisons (Marzuki et al., 2023).
Reference
Marzuki, Widiati, U., Rusdin, D., Darwin, & Indrawati, I. (2023). The impact of AI writing tools on the content and organization of students’ writing: EFL teachers’ perspective. Cogent Education, 10(2), 2236469. https://doi.org/10.1080/2331186X.2023.2236469
Useful questions
Additionally, you may consider the following factors when choosing your prompts.
- How do you intend AI to respond?
- What do you know about the AI model that you are working with?
- What specific domain does your task fall under?
- Are your instructions sufficiently clear and specific?
- What are the constraints that AI might have?
The quality of AI output largely depends on the amount of details provided by the user. While it can streamline your workflow, you should always consider the constraints of AI models and decide how to best utilize these tools.
Tips
- Experiment with different approaches. If the AI misunderstands, rephrase your prompt.
- Think critically—AI is prone to errors, so make sure to double-check its information to spot inaccuracies.
Designing effective questions or clear instructions for AI can help optimise the quality of the results. However, AI can misinterpret prompts and provide information that is false or outdated. AI response could also lack depth and lead to superficial understanding. Students should always critically verify the information provided and never over-rely on it.
