The Thompson Library & Archives Blog

Dean's Notes: Recap from Hart-Davidson Lecture

Posted on behalf of Dr. Jennifer Dean, Director of the Thompson Library.

Thanks to a timely notification from the TCLT, I had the opportunity to visit Mott Community College and attend a lecture by Dr. Bill Hart-Davidson of Michigan State University in October. It was my first visit to Mott since I began in Flint in 2022, and I had minimal information about where the lecture was being held. It was raining, I wasn’t sure where to park, and I felt nervous and out of my element. Fortunately, I stuck with it. Dr. Hart-Davidson’s talk was engaging, thought-provoking, and just plain helpful. I realized that my experience attending the lecture at Mott was not unlike the way we might feel as we learn more about AI. Though I have read and talked a lot about generative AI, I am still hearing a lot of confusion - well, frustration, really - when I talk with colleagues in the library and higher education worlds about AI. We know it’s here, we know it will affect us…but we’re just not sure how yet, and sometimes it feels like we have to talk about it when we’d really rather be doing something else. Dr. Hart-Davidson’s talk was just what I needed: good information and useful tools. I’d like to share some of my takeaways, in case they are helpful for you, too.

Dr. Hart-Davidson holds a PhD in Rhetoric & Composition from Purdue University and serves at MSU as Associate Dean for Graduate Education in the College of Arts & Letters; Professor in the Department of Writing, Rhetoric, & American Cultures; and Senior Researcher in the Writing in Digital Environments Research Center. He is a co-inventor of Eli Review, a software service that supports writing instruction and co-founder of Drawbridge, a learning technology company. His combination of education and experience make him uniquely qualified to help instruction colleagues understand and think about how to work with AI.

Some AI background I learned from Dr. Hart-Davidson’s lecture:

• AI is here, and it’s been here - we’ve been using it for a while.
• The latest version of GPT, or Generative Pre-training Transformer, that makes up ChatGPT, uses natural language so that we can talk with it. This is a new and useful development, and also why it feels like AI is just suddenly here.
• AI comes in a “wrapper” or tool that makes it useful. ChatGPT is AI in a chatbot wrapper that we can converse with.
• How can we help others identify AI embedded in other tools? We’re using AI in our email and word-processing software, and even in our cars. We will need to use our judgment to determine whether a tool is right for us, and whether the advice the tool’s version of AI is providing us is right for our situation.
• AI tools must be trained - the advice they give is based on their training.
• Training involves ingesting information. What is in this information? Who created it? How has it been evaluated?
• The Transformer, or T in GPT, is what gives ChatGPT its speed - it transforms words into tokens.
• ChatGPT doesn’t need to look at individual words - it looks at the tokens it has grouped words into, and just needs to choose the right combination of tokens to match a query. This is how ChatGPT can do what it does so quickly.
• For all its ability, ChatGPT hallucinates! It cheerfully gives wrong yet convincing information.

Dr. Hart-Davidson had good advice about using ChatGPT in academic spaces, including developing writing beyond simply idea generation and drafting. Review and revision are human activities, as is providing cultural information and context. Students will need to know when it is okay to use an AI tool for an assignment, when it isn’t, how they may use it, and how to disclose that they have used it. We must develop our processes and show our work, ensuring we have incorporated transparency.  To my way of thinking this is not unlike the research process we are already familiar with.

A final thought, from a library perspective: Focus on review includes the literature review. Almost certainly, much of the material that has been used to train AI tools came from library collections, which are made up of works created by scholars, artists, and creators using established standards and review processes. Understanding the literature review, research, and publication processes will be essential as we navigate new challenges and opportunities presented by AI and continue to deal with the ever-present threats of bias, misinformation, and disinformation. Our library team is here for you to help you determine the best information resources for your project, find and use information ethically, and verify and cite information properly. We’ll learn to navigate this AI age together.