Artificial Intelligence (AI) Guide

What is a prompt?

Prompts are questions or commands that users type into the dialog box of an AI platform. Outputs are AI generated responses to users' prompts.

How do AI Prompts Differ from Search Engines?

• Use complete sentences and natural language for AI prompts rather than relying on keywords
• Provide context such as describing the audience for the answer
• View the prompts as conversations and continually ask for the information you need

The CLEAR framework from Dr. Leo S. Lo, the Library School Dean at the University of New Mexico, is an excellent framework for constructing prompts that was introduced in his article, The CLEAR path: a framework for enhancing information literacy through prompt engineering

C: Concise

L: Logical

E: Explicit

A: Adaptive

R: Reflective

• Create prompts that are succinct, prioritize the most important information and avoid unnecessary words.
• Concise doesn’t mean the prompt has to be short!

• Structure information in a logical order
  • This is different than boolean operators in databases in which the order doesn’t matter
• Consider the background :and explore the connections

Bad Example:

I'm doing some research on inflation and I need all kinds of information. Can you please tell me about recent things that have been happening with inflation and the economy to make prices really high?”

Good Example:

List important factors in the United States  that have contributed to inflation over the last 3 years and when these pressures ease?

• Establish clear guidelines and structure.
• Determine the required length, reference materials, and specific vocabulary.
• Define the appropriate tone and readability level for the content
  • You can ask AI to take on different roles, such as a professor, a practitioner in the field, or a high school teacher, to present information or answer questions in the style most suited to your needs.
  • If you need a detailed answer, ask AI to respond as a college professor. For a simpler explanation, ask AI to explain it as if to a child
• Ask for an output style:
  • Formats such as a bulleted list, detailed explanations and 

Scenario: You are a business student who is working on a financial case study for Disney.

Bad Example:

Good Example:

• Try different approaches and compose follow-up prompts asking for more specific information
• Even skilled prompt writers often have to write follow-up prompts to get what they need

Example: I need low-cost marketing ideas for a real estate firm serving affluent customers.

After receiving limited information on social media, you can ask a follow-up prompt, “Can you provide more details on using social media to attract customers?”

• Think of ideas on how you can improve your prompts
• Examine the quality of the AI responses you receive

Example: Construct a prompt for marketing ideas for the library. 

Ask questions such as:

• How useful were the responses?
• Did the follow-ups prompts improve the information you received?
• How did the previous categories of CLEAR such as Explicit and Logical improve your answers?

How are article databases different from AI?:

• While the pro versions of Chat GPT and Gemini are improving in providing scholarly sources, they still lag significantly behind library databases and Google Scholar. 
• Some of the citations provided by AI may be fake, often referred to as “ghost citations” or “hallucinations.”
• AI is like a black box and there is a lack of transparency
• With prompts we’re not just looking for information, we are generating new information.

Best Practices for combining Library databases with AI:

• Ask AI to generate a list of keywords for searching the databases.
• Similarly, ask AI to list some of the important themes for a topic
• AI can effectively interpret complex scholarly articles that users located in Library databases, especially the results sections that contain extensive mathematical analyses.

Student’s research question: 

• “Why does the satisfaction of buying a luxury product die down once we own it?”
• None of the keywords in the students’ question yielded relevant articles

Chat GPT’s Response when asked about relevant keywords for this topic:

When searching for scholarly articles on why satisfaction decreases after purchasing luxury products, you might use the following keywords:

• “Hedonic adaptation luxury goods”
• “Consumer satisfaction luxury products”
• “Post-purchase dissonance luxury”

Searching the EBSCO databases with the term “hedonic adaptation” produced several useful articles.