Skip to main content
Changes ahead
Some pages will look different than others as we modernize to bring you an even better website.
Navigate to Homepage Mobile Menu

ChatGPT

  • What is ChatGPT?
  • Using ChatGPT Effectively
  • Fact-Checking
  • Citing Generative AI
  • Beyond ChatGPT

NOTE for users of screen readers: The collection of tabs above, while presented as a list, is clickable. Navigate to the list item you want, and press space to activate that tab.

What is ChatGPT?

ChatGPT is an artificial intelligence (AI) tool that uses natural language processing techniques to respond to user-generated prompts. The "GPT" initials stand for generative pretrained transformer.

Put simply: You ask ChatGPT a question or provide a prompt, it replies using natural language. The following is a sample ChatGPT prompt and response.

Prompt: Who are you?
Response: I am an artificial intelligence program designed to assist with answering questions and providing information on a wide range of topics. I do not have a physical form, but I am able to communicate with users through text or voice. Is there anything else you would like to know?

Three versions of ChatGPT are available: a free version and ChatGPT Plus and ChatGPT Enterprise, both available by paid subscription. Learn more at ChatGPT. Create an account at ChatGPT's Get Started screen.

What do you need to know about ChatGPT?

ChatGPT is not always trustworthy

  • It was trained using a massive dataset of text written by humans that was pulled from the internet.
  • Thus, the responses can reflect the biases of the humans who wrote the text used in the training dataset.
  • It isn't connected to the internet and the data used to train it was collected prior to 2021.

ChatGPT sometimes makes stuff up

  • To make up for knowledge gaps (e.g., lack of training data to pull information from), it will provide a response to the best of its ability (often made up) rather than say “error” or “cannot compute.” This is known as "hallucinating."

What is ChatGPT good for and not good for?

Remember, you'll always need to verify the information, because ChatGPT will sometimes make things up.

What is it good for?

  • Brainstorming ideas.
  • Narrowing your topic ideas for a research paper, and keywords for searching in library databases or web search engines.
  • Explaining information in ways that are easy to understand.
  • Summarizing and outlining.
  • Asking questions (be sure to fact check the results). You can ask a million questions without fear of being judged.
  • Translating text to different languages (not completely fluent in every language).
  • Helping write or debug computing code.

What is it not so good for?

  • Library research (not yet). ChatGPT makes up citations that don't exist. It might give you articles by an author that usually writes about your topic, or even identify a journal that published on your topic, but the title, pages numbers, and dates are completely fictional. This is because ChatGPT is not connected to the internet, so has no way of identifying actual sources. For now, it's best to use Discovery, library databases, Google Scholar, or other reputable sources. This may change in the future with more specialized search tools based on Large Language Models (LLMs).
  • Asking for information on current events. This is because ChatGPT isn't connected to the internet and relies on data collected prior to January 2022.
  • Asking for any information that would have dire consequences if it was incorrect (such as health, financial, legal advice, and so on). This is because of its tendency to sometimes make up answers, but still sound very confident.

Different courses will have different policies

Check with your instructor for each course to find out the policy on using ChatGPT and similar tools. The best time to talk with your instructor is before you begin your assignment to avoid needing to start over if generative AI isn't allowed.

Be specific in how you plan to use generative AI. Would you like to use ChatGPT to help brainstorm ideas or come up with a topic for your assignment? Are you using it to summarize or explain complex concepts? Or do you plan to use it for writing and editing? Be prepared that your instructor may approve some use cases but not others.

Have a plan for giving credit. APA Style, MLA Style, and Chicago Style all have guidelines for citing generative AI. Your instructor may also ask for an appendix that includes the prompts that you provided to ChatGPT or the full transcript of your interaction.

Tutorials on ChatGPT and Generative AI

To learn more, try the University of Arizona Library's tutorials about ChatGPT. They contain short videos (3 min. or less), and quiz questions for self-review of what you learned.

  1. What is Generative AI?
  2. How does ChatGPT work?
  3. How did ChatGPT come to be?
  4. How does ChatGPT try to prevent harmful use?
  5. Using ChatGPT effectively

An eBook is also available: ChatGPT for Dummies (2023).

Adapted from ChatGPT and Education, Center for Innovative Teaching and Learning, Northern Illinois University, licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License and Student Guide to ChatGPT, University of Arizona Libraries, licensed under a Creative Commons Attribution 4.0 International License.

Prompting

What is prompting?
Simply, it's what you type into the chat box.

 

The way you prompt makes a huge difference in the output that ChatGPT gives you. So it's worth learning some tips.

 

Always verify the information it gives you.
Think of ChatGPT as your personal intern. They need very specific instructions, and they need you to verify the information.

ChatGPT sometimes makes things up. That's because it's designed to write in a way that sounds like human writing. It's not designed to know facts.

Tips for writing effective prompts

  1. Give it some context or a role to play.
  2. Give it very detailed instructions, including how you would like the results formatted.
  3. Keep conversing and asking for changes. Ask it to revise the answer in various ways.

Examples

  1. A role could be, "Act as an expert in [fill in the blank]."
    Act as an expert community organizer.
    Act as a high school biology teacher.
    Act as a comedian.

  2. Example prompt:
    Act as an expert academic librarian. I’m writing a research paper for Sociology and I need help coming up with a topic. I’m interested in topics related to climate change. Please give me a list of 10 topic ideas related to climate change.

  3. Example of changes: (keep conversing until you get something useful)
    Now give me some sub-topics or research questions for [one of those topics]. And give me a list of keywords and phrases I can use to search for that topic in library databases and Google Scholar.
    Or...
    I didn't like any of those topics. Please give me 10 more.

For more tips, see Top 13 ChatGPT Prompt Engineering Hacks. If you want to learn more about prompting, try this free course: Learn Prompting.

More tips for ChatGPT

  1. Sometimes it gets confused if you change topics in the middle of a conversation. When you want to change the subject, start a new chat.

  2. It will remember what you've said in the course of a conversation, so you don't have to repeat everything again. Just continue like you're talking to your intern.

  3. Don't ask ChatGPT (free version) for a list of sources. It will make them up. Instead use the Library's Discovery search, library databases, or Google Scholar.

  4. Choose an output format. In addition to paragraphs it can give you a table, a bulleted list, ascii art, multiple choice quiz questions, emojis, computer code, and more.

  5. In ChatGPT you can see a history of your conversations and in the settings you can delete your history and turn off the saving of future history. You can also export your history and save it on your own computer.

  6. Remember, don't enter any personal, private data in ChatGPT, because OpenAI may use your input to help improve the model. The free version is a research experiment. If you don't want your data used to help improve ChatGPT, you can turn it off in the settings (which means it also won't save your previous chats for your own viewing).

Fact-checking is always needed

AI "hallucination"
The official term in the field of AI is "hallucination." This refers to the fact that AI tools sometimes "make stuff up."

Which models are less prone to this?
GPT-4 (the more capable model behind ChatGPT Plus and Bing Chat) has improved and is less prone to hallucination. According to OpenAI, it's "40% more likely to produce factual responses than GPT-3.5 on our internal evaluations." But it's still not perfect. So verification of the output is still needed.

ChatGPT often makes up fictional sources
One area where ChatGPT usually gives fictional answers is when asked to create a list of sources. See the X (formerly called Twitter) thread, "Why does chatGPT make up fake academic papers?" for a useful explanation of why this happens.

There is progress in making these models more truthful
However, there is progress in making these systems more truthful by grounding them in external sources of knowledge. Some examples are Bing Chat and Perplexity AI, which use internet search results to ground answers. However, the Internet sources used, could also contain misinformation or disinformation. But at least with Bing Chat and Perplexity you can link to the sources used to begin verification.

Scholarly sources as grounding
There are also systems that combine language models with scholarly sources. For example:

  • Elicit
    A research assistant using language models like GPT-3 to automate parts of researchers’ workflows. Currently, the main workflow in Elicit is Literature Review. If you ask a question, Elicit will show relevant papers and summaries of key information about those papers in an easy-to-use table.
  • Consensus
    A search engine that uses AI to search for and surface claims made in peer-reviewed research papers. Ask a plain English research question, and get word-for-word quotes from research papers related to your question. The source material used in Consensus comes from the Semantic Scholar database, which includes over 200M papers across all domains of science.

How to cite generative AI

Remember to check with your instructor. Some instructors might not allow any use of ChatGPT and others might allow only limited use.

If you are allowed to use ChatGPT in an academic assignment, here are some guidelines for citing.

Your instructor may also ask for an appendix that includes the prompts that you provided to ChatGPT or the full transcript of your interaction.

For guidelines on citing other formats of generative AI (images, code, videos, etc), see How to Cite AI Tools: A Guide for Students.

See also this summary of ways to acknowledge use of generative AI: Acknowledging and Citing Generative AI in Academic Work.

Beyond ChatGPT: other useful language models

Copilot (Microsoft)
Copilot is built on the free version of ChatGPT, but unlike ChatGPT it's connected to the internet, meaning it can provide up-to-date information. It also links to the websites it gets its information from.

Gemini (Google)
Gemini is also linked to the internet. It includes links to websites with "similar content" in its results, but they aren't necessarily what it used to generate the results.

For a summary of some other AI tools, see the University of Arizona Library.

More extensive lists are available at AI Tools Landscape and Future Tools.