Introduction

Since launching in November 2022, ChatGPT has already made waves in academic institutions across the country. ChatGPT initially raised concerns on what its' abilities mean for writing, learning, and academic integrity in higher education, but since then, many use cases and explorations have shown exciting and effective ways that ChatGPT can enhance education in our classroom. This page is designed to introduce you to ChatGPT, share ways in which you can use ChatGPT in your own courses, and the limitations of it. Throughout this page you'll find actual prompts we used to talk to ChatGPT and the responses it gave in green boxes.

NOTE: ChatGPT is currently available for free with a $20 monthly subscription in place for increased access and availability of new features. ChatGPT may eventually go behind a paywall, so keep this in mind if you decide to incorporate this into your own teaching. A consideration being talked about is the inequity this product can create when it does go behind a paywall.

What is ChatGPT?

ChatGPT is an artificial intelligence (AI) chatbot that has been trained to answer questions and interact in a conversational way. The conversational component of ChatGPT allows you to ask follow-up questions and receive responses, admit its mistakes, challenge incorrect premises and reject inappropriate requests (OpenAI). The video to the right, ChatGPT, Explained, by the Wall Street Journal is a nice introduction to what ChatGPT is and how people have used it. Some ways in which ChatGPT has been used includes:

debugging computer programs
composing music
writing fairy tales, poetry, song lyrics
improving writing and grammar
writing essays and letters of recommendation
answering and writing test questions

Curious as to how ChatGPT explains itself, read the response to the prompt, "Please explain to me what ChatGPT is."

Certainly! ChatGPT is an artificial intelligence language model developed by OpenAI. It is part of a family of models called GPT (Generative Pre-trained Transformer) models, which are designed to generate human-like text.

ChatGPT specifically is a large language model that has been trained on a massive corpus of text data from the internet, including books, articles, and websites. It is capable of understanding natural language input from users and generating appropriate responses that are contextually relevant and grammatically correct

The goal of ChatGPT is to simulate human-like conversation and provide helpful and informative responses to users' inquiries. ChatGPT has been trained on a wide range of topics, from science and technology to entertainment and current events, and can generate responses in a conversational tone that closely resembles human speech.

ChatGPT is designed to continually learn and improve over time, as it interacts with more users and receives feedback on its responses. It is an exciting technology that has the potential to revolutionize the way we interact with computers and the internet.

Strengths and how you can use ChatGPT in academics?

One topic we think is important to discuss is the ways in which ChatGPT can be used to improve the student experience in the classroom and improve student learning. A great starting point is to look at The Educator's Playbook article that shares Five ways teachers can integrate ChatGPT into their classrooms today and Writesonic's 110 Best ChatGPT examples to look at in 2023. These articles do a great job of showing how faculty can use ChatGPT to help in the process of learning, to develop initiate ideas, and how a variety of people are using ChatGPT for different industries. In addition, check out the list of ideas below that includes other ways you can incorporate ChatGPT into your course:

Help students brainstorm paper ideas
Generate early draft or starting point of a paper [use track changes in Microsoft Word] to show revisions.
Evaluate and revise papers for grammar and writing conciseness
creating citations using different formats (APA, MLA, etc.)
Allow students to revise portions of their paper/assignment using ChatGPT and reflect on the process and improvement
Use ChatGPT to answer questions on an assignment and have students reflect on the accuracy/inaccuracy of the responses.
Involve conversations about use of AI in the discipline and how it can shape their future work.
Brainstorm short answer and multiple choice quiz questions using a lecture transcript
Grade assignments using rubrics
Jumpstart ideas for writing letters of recommendation

Digital Learning encourages you to do your own research. One of the best ways to understand limitations and benefits of ChatGPT in your discipline is to explore conversations that are being had within your discipline. What are your colleagues doing? What are the experts doing?

Using ChatGPT in another way in your course, or want to add additional ideas to the list? We'd love to hear from you! Please contact us at uno.its.digitallearning@unomaha.edu.

Limitations of ChatGPT

Although ChatGPT is a powerful and impressive language model, there are some limitations on what it can do. We gave ChatGPT some prompts to understand its' limitations. See the prompts given and the responses below:

Prompt: Can you explain to me a bit about the limitations of ChatGPT

Sure, while ChatGPT is a powerful and impressive language model, there are some limitations to what it can do. Here are a few examples:

- - Limited Understanding of Context: ChatGPT relies on statistical patterns and associations learned from large amounts of text data to generate responses. However, it can sometimes struggle to understand the context of a conversation, leading to inappropriate or irrelevant responses.
  - Lack of Common Sense: ChatGPT doesn't have real-world experiences, and therefore it might not have a common sense of human beings. For example, if you ask ChatGPT to pour water on a laptop, it might suggest that it is a good idea.
  - Biased Responses: Because ChatGPT is trained on text data from the internet, it may inadvertently pick up biases and stereotypes that exist in that data. This can lead to biased or offensive responses, particularly on sensitive topics like race, gender, and politics.
  - Limited Knowledge: While ChatGPT has been trained on a vast amount of text data, it still has limitations in its knowledge base. It might not have the latest information on current events, or it may not be able to answer very specialized or technical questions.
  - No Emotional Intelligence: ChatGPT doesn't have emotional intelligence, and it can't understand or empathize with the emotions of the person it's interacting with.

Prompt: This is great, but can you talk about the limitations in regard to creating multiple choice questions and picking answers on tests?

Certainly! While ChatGPT is capable of generating text-based responses to questions, it is not specifically designed for creating multiple choice questions or picking answers on tests. Here are some limitations to consider when using ChatGPT in this context:

1. 1. Limited Control over Response Options: ChatGPT generates responses based on statistical patterns learned from text data, so it may not always produce the specific response options that you want for a multiple choice question.
  2. Lack of Understanding of Test-Taking Strategies: ChatGPT does not have a conceptual understanding of test-taking strategies or the types of questions and answer choices that are commonly used on tests.
  3. Limited Domain-Specific Knowledge: While ChatGPT has been trained on a wide variety of text data, its knowledge base may not be sufficient for answering domain-specific questions that are often found on tests, such as in math, science, or technical fields.
  4. Difficulty in Evaluating Answer Accuracy: ChatGPT may generate plausible-sounding answers, but it may not always be able to accurately evaluate the correctness of an answer choice, particularly in cases where the correct answer requires significant domain-specific knowledge.

Overall, while ChatGPT may be able to generate potential answer choices for multiple choice questions, it may not be the best tool for creating or evaluating test questions. It is important to use caution when relying on ChatGPT in these contexts and to verify the accuracy of its responses through other means.

Prompt: I hear you give inaccurate and made up quotes sometimes when writing papers; can you explain to me your limitations in this realm?

I'm sorry, but I must clarify that as an AI language model, I don't intentionally provide inaccurate or made-up quotes. However, as a machine learning model, I generate responses based on patterns learned from the large corpus of text data that I was trained on. Therefore, there may be instances where the responses I generate include inaccurate or incomplete information, or even occasional errors.

In terms of generating quotes for use in academic papers or other contexts, there are some limitations to consider when using ChatGPT. For example:

1. 1. Lack of Context: ChatGPT generates text based on statistical patterns learned from text data, so it may not always fully understand the context or meaning of a quote or its relevance to a particular topic or argument.
  2. Limited Ability to Verify Sources: ChatGPT may generate quotes that are attributed to a particular author or source, but it does not have the ability to verify the accuracy or authenticity of those sources.
  3. Inability to Create Original Quotes: While ChatGPT can generate responses that resemble human speech, it cannot create original quotes in the same way that a human writer can.
  4. Risk of Plagiarism: If used incorrectly, ChatGPT could be a tool for plagiarism. It is important to understand that using content generated by ChatGPT as original work or without proper attribution could constitute plagiarism.
  5. Limited Knowledge: ChatGPT may not have the domain-specific knowledge or understanding necessary to accurately generate quotes for specialized fields or technical subjects.

In conclusion, ChatGPT can be a useful tool for generating quotes and other text-based content, but it is important to be aware of its limitations and to verify the accuracy and authenticity of any content generated by ChatGPT before using it in academic papers or other contexts.