Prompt Engineering
Prompt engineering is the method to ask generative AI to produce what the individual needs. There are two main principles of building a successful prompt for any AI, specificity and iteration. The box below includes one example of a framework that can be applied when prompting any generative AI.
AI Prompt Frameworks
The "CLEAR" framework to construct prompts:
| Concise | Brevity and clarity in prompts by removing superfluous language from a prompt so AI can focus on key components. Example: "Explain entropy and its significance" versus "Please provide a detailed explanation about entropy and its significance." |
| Logical | Structured and coherent prompts to create a logical flow and order of ideas. Example: "Take me through the steps of building a good prompt, starting with how to develop a topic or question to retrieving the prompt result." |
| Explicit | Clear output specifications that provide precise details about output format, content or scope. Example: "Provide a 25 minute podcast script that provides a concise summary of entropy, its discovery and impact." |
| Adaptive | Flexibility and customization in prompts to allow refinement of an initial prompt. Example: "Provide a 25 minute podcast script that provides a concise summary of entropy, its discovery and impact - gave too vague of results, an adaptive prompt might be, how does the concept of entropy apply in biology and apoptosis." |
| Reflective | Continuous evaluation and improvement of prompts to finally retrieve the results that will be useful to you. |
Source: Lo, Leo, S. (2023). The CLEAR path: A Framework for Enhancing Information Literacy through Prompt Engineering. The Journal of Academic Librarianship, 49(4).
| Subject |
Describe the subject in as much detail as you can |
| Style | Specify the type of media or the specific style of art |
| Details | Include further specifications about color, lighting, and so forth |
| Format of Output | Specify the format the image will be, i.e. poster |
Source: Harvard University, Information Technology, Getting Started with Prompts for Image-based Generative AI Tools, November 16, 2023
Key Takeaways
-
Be as specific as possible.
-
Most AI tools are not able to interpret abstract or nuanced concepts.
-
Use sequential prompting.
-
Each subsequent prompt is informed by the AI's previous response, creating a chain of interactions that progressively refines the model's output.
-
Provide examples for the AI tool to imitate.
-
Role-play with real-world scenarios and personas to provide context clues.
-
Use ‘Do’ or ‘Don’t’ commands.
-
Tailor your approach with the AI tool’s capabilities and limitations in mind.
Learn More about Frameworks for Prompt Engineering
- AI regulation guide: Using PICO to Generate Evidence for AI Development (2021, November 9). NHS Transformation Directorate.
- Awais Ahmed, Xiaoyang Zeng, Rui Xi, Mengshu Hou, Syed Attique Shah, MED-Prompt: A Novel Prompt Engineering Framework for Medicine Prediction on Free-text Clinical Notes, Journal of King Saud University - Computer and Information Sciences, Volume 36, Issue 2, 2024, 101933.
- Blasingame, Mallory N., et al. "Evaluating a Large Language Model's Ability to Answer Clinicians' Requests for Evidence Summaries." medRxiv (2024): 2024-05.
- Gattupalli, S., Maloy, R., & Edwards, Sharon. (n.d.). Prompt Literacy: A Pivotal Educational Skill in the Age of AI.
- Jacobs, H. H., & Fisher, Michael. (2023, June 26). Prompt Literacy: A Key for AI-Based Learning. Educational Leadership Magazine, 80(9).
- Kominato, Hero. Timeless Prompt Engineering for AI Image Generation. Medium, Sept. 27, 2023.
- Liu, Vivian, and Lydia B. Chilton. "Design Guidelines for Prompt Engineering Text-to-image Generative Models." Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems. 2022.
-
Oppenlaender, Jonas. A Taxonomy of Prompt Modifiers for Text-to-image Generation. Behaviour & Information Technology (2023): 1-14.
-
Scharth, Marcel. (2023, February 23). How to Perfect your Prompt Writing for ChatGPT, Midjourney and Other AI Generators. The Conversation.
How to Access the Full Text of an Article
"Get it at UC" Button
Most databases provide a description of articles and other resources to help you find what you need by searching through the indexed literature within their website. Included as part of the item record is the “Get it at UC” button. Clicking this button will direct you to a copy of the full text of the article.
Interlibrary Loan Request
If we do not own a journal or book, you can submit an interlibrary loan (ILL) request to have the book or article (e)mailed to you for free from another UC library. Learn more about how to request books or articles.