Feature Stories Campus Events All Stories

W&L Dialogues on the Future of AI Faculty and staff dive into conversation and collaboration in response to ChatGPT and other emerging technologies.

DSC08026k-800x533 W&L Dialogues on the Future of AIPaul Hanstedt, director of W&L’s Harte Center for Teaching and Learning, explains ChatGPT’s capabilities and limitations.

Northen Auditorium in Washington and Lee’s Leyburn Library was standing room only at a recent talk about the future of artificial intelligence (AI) and its role in the higher education landscape on Thursday, Jan. 26.

“We’re at an inflection point in an exponential growth curve,” said presenter Jeff Schatten, associate professor of business administration, of AI’s rapid development within the past year.

Schatten’s talk, “The Future Is Now: ChatGPT and AI,” was offered as part of ITS Academic Technologies’ “Technology and Tacos” series. This ongoing series hosts presentations for the W&L community that provide learning opportunities and prompt discussion on technology that impacts student learning.

Academic Technologies and the Harte Center for Teaching and Learning partnered to offer W&L faculty and staff on two related presentations in January. The week following Schatten’s presentation, Paul Hanstedt, Director of Houston H. Harte Center for Teaching and Learning, partnered with Academic Technologies to offer a workshop called “The Best Response to ChatGPT and AI? A Solid Pedagogical Practice.” Both presentations received some of the highest attendance numbers that the series has seen since its inception.

In September 2022, Schatten wrote an op-ed for the Chronicle of Higher Education about ChatGPT, an artificial intelligence language model. He has closely followed the progress of ChatGPT, which stands for Generative Pre-trained Transformer, as it begins to gain traction on college campuses as well as in business, scientific research and other areas. (Schatten even wrote a children’s book using ChatGPT.)

Schatten’s presentation provided a brief overview of the current state of AI, including tools like ChatGPT that create imagery, music and even mimic the human voice. Schatten’s presentation also demonstrated more recent developments such as the cross-pollination of what ChaptGPT can do alongside an AI image generator and played a piece of AI-generated music. He said that the technology’s recent strides are only the beginning of its capabilities and reach.

ChatGPT was developed by OpenAI, a leading research organization focused on advancing artificial intelligence. It uses a type of machine learning called deep learning, which involves training a neural network on large amounts of data, pulled from the web, books and other sources. Since its initial release in 2018, ChatGPT has been updated and refined several times, with the latest version being particularly notable for its impressive language capabilities and versatility. ChatGPT can be utilized in an academic setting in several ways, including asking questions about coursework and assignments, or asking for specific feedback on written work. The model can also assist with research tasks by providing writing support, generating citations, answering questions related to specific fields of study, and even answering questions in a conversational tone that mimics human responses.

DSC08056k-800x533 W&L Dialogues on the Future of AIFaculty gathered in the Harte Center for Hanstedt’s presentation.

There are aspects of the model that are cause for concern. One issue is the model’s potential for racial or gender bias, despite attempts by its developers at data cleaning and other techniques for retraining it. Another concern is the potential for the model to generate false or misleading information; one example in Schatten’s presentation was a research paper that cited an entire study that was never actually completed.

There are also concerns about the ethical implications of using language models like ChatGPT for tasks such as writing assistance, and colleges and universities are currently grappling with the potential negative consequences for learning outcomes or the quality of written work as students become more familiar with AI as a tool.

Hanstedt’s presentation on Jan. 31 began by asking faculty to approach addressing ChatGPT and other AI tools with students as a learning opportunity grounded in trust.

“We have good, thoughtful, hardworking students,” Hanstedt said, “and we need to begin by walking into our classrooms and assuming the best.”

Hanstedt’s presentation highlighted ChatGPT’s current limitations, given that its training relies on predictive text and large data sets, and focused on ways to help students understand the benefits of process-oriented, experiential assignments that are geared toward meaning-making.

“The model doesn’t truly understand the meaning when it’s putting words together,” Hanstedt said. “It is simply predicting, based upon this huge dataset, what words should come next.”

Hanstedt noted that video projects, podcast assignments, and research that requires localized community data and interviews can engage students with the material in a way that cannot be pulled from existent data sets or otherwise duplicated by AI.

Some faculty have already incorporated ChatGPT into assignments and are engaging in conversation with students about AI’s applicability to their coursework.

“We’re preparing students for professional careers — this is something they will need to know how to use,” said Gavin Fox, associate professor of business administration. “I encourage my students to learn how it works and how to use it.”

DSC08046k-800x533 W&L Dialogues on the Future of AIFaculty had the opportunity to ask questions and work in small groups to brainstorm approaches to working with AI in the classroom.

Julie Knudson, director of Academic Technologies, said that faculty began reaching out prior to organizing the January sessions to start a dialogue on campus about how to approach new AI developments, particularly in creating writing assignments.

“There are many ways that you can approach your classes to make them local, make them specific, and make them based on experiential learning,” Knudson said. “A lot of the ways that we already encourage professors to incorporate technology into their classes, like using digital storytelling, podcasting, and incorporating film or video into their teaching, renders ChatGPT less useful.”

Knudson said that the resources like video editing suites and equipment for quality audio recording are available through Academic Technologies and the Harte Center. Harte Center faculty teaching fellows, who are available to collaborate with faculty on pedagogical approaches to their courses, make the university well-positioned to support faculty in finding ways to deepen student learning through more experiential and process-oriented assignments.

“We have always had a culture of faculty participating in workshops like these and continually developing their teaching, which has only expanded with the Harte Center,” Knudson said. “So we are in a very good place as far as being able to hold workshops led by folks who are dedicated to thinking about pedagogy, in great facilities — we’re ideally positioned for engaging with this.”