Administration and faculty at Salt Lake Community College are gearing up to address ChatGPT and other artificial intelligence programs that provide detailed responses to prompts.
Four internal talks taking place this month aim to explore the potential benefits and downsides of the technology, such as facilitating research and plagiarism. SLCC Dean of Students Candida Mumford said the question comes down to how to juggle integrity and innovation.
“How do we uphold academic integrity while allowing these innovations to happen in the classroom? We’re not the only institution grappling with this,” Mumford said.
Since the public release of ChatGPT in November 2022, users have used the program to complete assignments with minimal personal input. On March 14, developer OpenAI announced the latest version of its program, GPT-4, which the developer claims can clear the uniform bar exam and land in the top 10th percentile.
ChatGPT accomplishes this as a model trained on large amounts of data pulled from the web. For this reason, Business Professor Dr. Lon Schiffbauer prefers to call the program “informed” rather than “intelligent” for the time being.
“It’s informed by the data that it has brought together and creates an autonomous output based on [that],” Schiffbauer said. “But if you’re informed about bad knowledge, it’s still an informed autonomous decision, but … not intelligent.”
Clint Gardner, director for the Student Writing and Reading Center – which provides tutoring sessions to help students with writing assignments – pointed to the technology’s over reliance on “token phrases” when assembling text. In other words, language that is redundant.
“It’s very factually challenged at times,” Gardner said. “[People] are more reliable than a prediction engine.”
The U.S. Department of Education maintains that artificial intelligence, or AI – a broad term used to describe emerging machine learning technologies like ChatGPT – will have “powerful impacts on learning” through support for students and by “empowering educators to be … less consumed by routine, repetitive tasks.”
This prospect excites Schiffbauer as well as English Professor Tiffany Rousculp, the latter of whom said they believe ChatGPT can redirect learning efforts for the better. Both co-directed a workshop covering the program during this year’s SLCC 360, an event that updates staff and faculty on the college’s trajectory.
“There’s an opportunity for more engaging assignments than the standard research paper,” Rousculp said. “It’s not a problem to be solved, it’s a new technology to be incorporated into student learning.”
Rousculp went on to say that SLCC administration and faculty agree the majority of students want to learn and believe misuse of ChatGPT will only occur among a small percentage. Still, Dean Mumford said an update to the code of student rights and responsibilities is forthcoming.
“As far as the code,” Mumford said, “[ChatGPT misuse] incidents will be navigated much like our other academic misconduct. It’s considered presenting someone else’s ideas that are not your own.”
GPTZero, software that helps detect AI-generated text, has partnered with Canvas, the application SLCC uses for its web-based learning system. Although Canvas has yet to integrate the tool, Rousculp said she believes that detection shouldn’t be the end-all solution.
“We [must] have ways that students use this to their benefit and learn … using it in their eventual profession, whatever field they go into,” Rousculp said. “We can’t detect and forget.”
Mumford said everyone in academic settings has an obligation to find ways to move forward with ChatGPT.
“Let’s teach ourselves and our students how to use it in a responsible way,” Mumford said.