One of the recommended ways to use AI in research is prompt engineering. This means that you can get better answers from AI if you know how to ask a question in the correct way. Often, to get the best AI results, you have to do a lot of the work at the beginning yourself. For example, if you need help generating good research questions, you'll need to work on your topic first and be sure it is specific enough to get excellent, rigorous research questions from the AI Tool.
A prompt is (technically speaking) input to an AI tool. For research AI tools, the prompt is almost always a question, request, or topic posed by you, the human researcher. Why should you try to write good prompts?
The CLEAR Framework, designed by librarian, Dr. Lo of the University of Arizona, provides "a standard method for composing effective queries for AI language models" (p.1)
| Good | Needs Improvement | |
|---|---|---|
| Concise: Brevity and clarity in prompts | Explore the impact of AI on job automation in the healthcare industry, focusing on recent developments and implications. | I need information on jobs in the healthcare industry. |
| Logical: structured and coherent prompts | Generate research questions related to the impact of artificial intelligence on workforce automation in the healthcare industry. | Explain how AI is affecting the healthcare industry. |
| Explicit: clear output specifications | Identify peer-reviewed articles published within the last five years that examine the relationship between social media usage and mental health among adolescents. Focus on studies conducted in Western countries and include factors such as frequency of social media use, types of platforms used, and specific mental health outcomes measured (e.g., depression, anxiety). Exclude articles that are not written in English or are not available in full-text format. | Find a survey determining whether social media usage and mental health among adolescents will increase the minimum wage. |
| Adaptive: flexibility and customization in prompts | Focusing on developing countries, examine the effect artificial intelligence has on job displacement in the manufacturing industry. | Research artificial intelligence and the manufacturing industry. |
| Reflective: continuous evaluation and improvement of prompts | Reflect on the ethical implications of using genetic engineering to modify crops for increased yield in developing countries. | Reflect on the history of genetic engineering on farming. |
These examples were created with the help of Chat GPT 3.5
Using free AI tools (such as ChatGPT 3.5) is discouraged when you ask it directly to find sources. It will generate fake sources that look legitimate, but when you search for the full text, you'll find that it doesn't exist. Other AI Tools that require a paid subscription (such as ChatGPT 4.0) generate real peer-reviewed sources when asked. Because of this, if you'd like to use the free option, there are some steps you can take to generate the best key words in order to do a library search in the Libraries' database, Google Scholar, and others.
| Example | Key Words | Results |
|---|---|---|
| Efficacy of Novel Drug Delivery Systems in Targeting Cancer Cells | "novel drug delivery system" AND "cancer cells" | UofM Libraries Database Results |
| Cybersecurity Threat Detection using Machine Learning Algorithm | "cybersecurity threat detection" AND "machine learning algorithm" | UofM Libraries Database Results |
| Assessing the Impact of Climate Change on Coral Reef Ecosystems | "climate change" OR "global warming" AND "coral reefs" OR "bleaching" | UofM Libraries Database Results |
These examples were created with the help of Chat GPT 3.5
Georgetown University. Artificial Intelligence (Generative) Resources. https://guides.library.georgetown.edu/c.php?g=1352831&p=9985827
Lo, Leo S. (2023). The CLEAR path: A framework for enhancing information literacy through prompt engineering. The Journal of Academic Librarianship, 49(4), 102720. https://doi.org/10.1016/j.acalib.2023.102720