K-12 schools, universities in Austin adapting to viral AI chatbot ChatGPT

Although the artificial intelligence technology ChatGPT has been available to the public for only a few months, Central Texas school officials are already dealing with changes the tool could bring to education and how students learn.

ChatGPT, which launched in November, is an AI chatbot that can generate conversational, original responses to various questions. Anyone can enter a prompt for free, such as “Write a five-paragraph essay about the themes of 'The Great Gatsby,' " and the chatbot will generate a response, although the answers are not always accurate.

Since its launch, the chatbot has gone viral, prompting concern among some educators about how students could use it to cheat. However, others in academia say the chatbot, if used correctly, can be a resource to enhance how students learn about writing and editing in the classroom.

In late January, OpenAI, which created ChatGPT, launched a tool called the AI Text Classifier, which can distinguish between AI-written and human-written text. The creators, however, warned that the classifier isn't fully reliable and can incorrectly label human-generated text as written by AI.

Officials at K-12 schools and higher education institutions in Austin told the Statesman they are beginning to consider how to adapt their instructional methods and assignments due to the chatbot's launch, though they said that much of their response is still in the early stages because of the tool's novelty.

Writing instruction opportunities

Many K-12 school districts told the American-Statesman that even if officials are mulling how to prepare for the tool's use, they haven't yet developed any policies for it.

In the Austin district, students don't have access to ChatGPT on district networks, said Laura Browder, director of technology integration. Teachers have a variety of tools to check an assignment for plagiarism, such as those created by Google Assignments and McGraw-Hill, she said.

The Eanes district has begun developing ideas about how to prevent ChatGPT-related cheating, said Molly May, assistant superintendent of curriculum, instruction and assessment.

"We have investigated AI technology to mitigate cheating but have not found a tool that completely detects cheating," May said.

Some researchers are working on applications to detect AI-produced text, but in these early stages, it's probably not too hard to tell the difference between a chatbot and a student, said Catherine Lammert, an assistant professor in Texas Tech University's teacher education department. ChatGPT and other similar tools are impressive but do make errors, she said.

"The errors that it makes are going to be different than the errors a typical middle or high school student would make," Lammert said.

Instead, teachers could think about how to integrate the technology into their classrooms, said Peter Salib, assistant professor of law at the University of Houston Law Center. ChatGPT is often a good starting point, especially for students who struggle with writing, he said.

Teachers could assign work that encourages students to integrate ChatGPT into their writing or use it as a starting point, much in the way students sometimes use the online encyclopedia Wikipedia as a starting point to find other sources in research, Salib said

"It turned out no one could tell if you'd gone to Wikipedia and used that as your starting point because the sources were good," Salib said.

In the best-case scenario, using AI language models could encourage educators to rethink how they teach writing and critical thinking beyond the basic five-paragraph format, Lammert said.

Local colleges, universities respond

At the University of Texas, the Center for Teaching and Learning has shared optional recommendations for UT instructors on ChatGPT, which includes discussing the ethical implications of the chatbot, asking students to cite sources on assignments, and focusing assignments on skills that AI struggles with, such as leadership, collaboration or applying ethics or morals to a situation.

Art Markman, who co-chairs UT’s new ChatGPT working group, said the group will be focused on raising awareness of the tool, providing resources to help detect the tool's use on assignments and helping develop creative ways to use ChatGPT in the classroom, such as editing an essay or prompt produced by AI.

“This is a tool that has some functions that may be valuable for people, and we want to make sure that if it's a tool that's going to be adopted, that it becomes a tool that our students learn how to use,” said Markman, who is also UT’s vice provost for academic affairs.

Gaye Lynn Scott, vice chancellor of instruction at Austin Community College, said faculty members are considering different approaches for how to use ChatGPT, such as critiquing an essay produced by the AI. Both Markman and Scott said a campuswide ban is probably not on the table for either school, and even a ban that an instructor might institute within a class would be difficult to enforce.

“The opportunity here, for us as an institution and for our faculty, is to remember that we're trying to cultivate dispositions and characteristics in our students,” Scott said. “We're not just trying to cultivate mechanical ability. We're really trying to figure out how to help students understand how to do things and how to assess new tools.”

In separate interviews, Markman and Scott both compared the launch of ChatGPT to the introduction of a graphing calculator into classrooms. They said the calculator prompted math instructors to adapt and consider how to incorporate the tool into their instruction, and ChatGPT will probably motivate faculty members who teach subjects such as English to do the same.

“Every tool that's been created has ultimately enhanced the way that we teach,” Markman said. “It frightens us at first, and then it enhances it. When calculators first became widely available, people were afraid no one was ever going to learn basic math skills, and they still do. Calculators do enhance our ability to teach more complex math. These AI tools are going to be the same way.”

More likely than not, Salib said, educators at some point will need to reconcile with the technology.

"These things are not going away," Salib said. "They're not going to be worse. They're only going to be better. It seems really important that students learn how to use them."

This article originally appeared on Austin American-Statesman: How K-12 schools, universities in Austin are responding to ChatGPT