Skip to Main Content

Artificial Intelligence (AI): Home

What is AI? And what is an LLM?

A robot stands before a blackboard with mathematical equations.

Artificial intelligence is the application of computing power to create new combinations of text, images, and data analyses for people to use. In a pure vision of artificial intelligence, the computer would be replicating and possibly improving on the process of human thought. In the existing models, AI tools are programmed to make best guesses at what responses best fit the questions they receive. 

Generative AI is a type of AI that includes the chat features present in most of the tools in this guide. These generative tools required a Large Language Model (LLM) to learn how to generate text responses to prompts entered by users. The LLMs will vary in content and size, but may include all sorts of documents and publications.

A more detailed overview on these concepts is available in A Generative AI Primer from the Jisc National Centre for AI.


Questions about using AI? Contact Jessie Long ( or John Burke ( by email, use our chat service, or set up an appointment for a Research Consultation. You may also make appointments for Research Consultations in Navigate with Jessie Long and John Burke


[Image] "Artificial Intelligence & AI & Machine Learning" by mikemacmarketing is licensed under CC BY 2.0.

Where to start searching: general purpose AI tools

These AI tools all work quite similarly: start up a chat session with the AI, ask it for information, and review the results. Beyond these tools and the others on the tabs to the left, you can search through thousands of special purpose AI tools at There's an AI for That (TAAFT).

How to Write a Prompt

How you describe what you are searching for impacts your results, whether searching Google, a library database, or an AI tool.

Here are six strategies provided by OpenAI, the creators of ChatGPT. Briefly, they include these tactics:

  1. Provide details to get more relevant answers ("Tell me how to start looking for information on legal cases. I need to use recent legal cases that involve class-action lawsuits against corporations. Provide me with at least three options.")
  2. Ask the tool to adopt a persona ("Writing as a librarian . . ." or "writing for people unfamiliar with X . . . ")
  3. Provide a reference text to guide your answers ("Using the information in the three provided PDFs, write three short paragraphs that explain . . . ")
  4. Specify the steps required to complete a task ("First, create a list of possible search terms for this topic. Next, summarize the Wikipedia entry on the topic, listing words or phrases that seem important. Then . . . "). Be sure to break up lengthy queries (with many steps) into smaller ones.
  5. Let the AI think for a while about your question. Ask it to figure out an answer, and then have it compare that answer to one given in a source you provide, or have it answer the same question again and then compare the responses. 
  6. Specify the desired length of the output. ("Make a list of ten ways that . . . ")

You can also try a similar model for prompt engineering in the CLEAR method. The linked article outlines five core principles of prompt engineering:

  1. Concise - make clear what you are asking the AI tool ("Suggest four strategies . . . ")
  2. Logical - explain what you are looking for in a structured request (Describe X, first investigating Y, then looking at Z, and finally . . . ")
  3. Explicit - clearly provide the output you expect ("Create an analysis of X that includes its cause, the main events, and the outcomes in no more than 1500 words.")
  4. Adaptive - be ready to try alternative prompts if your first one does not work well (for instance, making your question more specific or clearly stating that it should involve a particular location or group of people).
  5. Reflective - carefully examine the response to see if it look accurate and ask the AI to provide more information or more information specific to what you asked.

Miami Policies on AI

Benefits and Drawbacks of Generative AI

  • Write text at various levels of complexity and meaningfulness on any topic.
  • Provide a summary of a longer document (either by pasting the text into the AI tool or uploading a PDF or other file).
  • Helps with brainstorming ideas and search terms for a topic.
  • Create AI-generated art.
  • Find answers to questions formed from a variety of sources from the tool's LLM.
  • Creates unique text each time it is used.
  • Hallucinations (can create text that sounds right, but is actually incorrect)
  • Privacy concerns (it uses the data you enter to train the model, so it can be shared with others).
  • May require you to set up a login and share personal information in order to use it.
  • Biases in the materials used to train it can flow through into its responses.
  • Different versions may have source or date limitations that make its answers incorrect or less useful.
  • Given the variety of free and pro versions, there can be digital inequity in access to the same level of tool.