Artificial Intelligence for Research and Teaching

Dr. Leo S. Lo, Dean of Libraries at the University of New Mexico, developed the CLEAR method for constructing and optimizing queries (prompts) in order to use AI language models (LMs) efficiently. His premise is that a well-defined prompt produces a more “meaningful and informative response, whereas  a poorly constructed one may result in outputs that are irrelevant or nonsensical.”

Lo, L.S. (2023, July). The CLEAR path: A framework for enhancing information literacy through prompt engineering. The Journal of Academic Librarianship. 49 (4). https://doi.org/10.1016/j.acalib.2023.102720

Five components of the CLEAR Framework:

Concise – brevity and clarity in prompts

• Instead of requesting: Can you provide me with the major sources of inflation in the 1970s and the impact it had on the American economy?
• Use a more concise and explicit prompt: Identify the leading causes of inflation during the Carter administration.

Logical – structured and coherent

• A structured prompt asks for steps in a process or procedure.
• Example: Construct the steps in setting up a gallery exhibit, from developing a theme, to working with the artists, displaying, marketing and promoting the exhibits, and planning the opening event.

Explicit – clear output specifications

• Instead of prompting: What is global warming?
• Use more precise instructions: Outline the effects of climate change on the planet emphasizing its major causes, both human and natural, and consequences.

Adaptive – flexibility and customization in prompts - "If at first you don't succeed, try, try again."

• First try: Examine the impact of video games on developing brains.
• Second try: Illustrate how video games may be linked to negative health outcomes in adolescents.
• Third try: What are the determinants for healthy video game use in adolescents?

Reflective – continuous evaluation, improvement of prompts … and further research

After receiving AI-generated responses about the effect of video games on adolescents, it was time to evaluate the answers and decide in which direction continued research on the subject should go. Given the overlap in the three responses from ChatGPT, it was evident that individual, social, and environmental determinants are at play in video game use by adolescents. Using library and other information sources to back up your hypothesis, the real research begins.