by Carlina Grillo

In recent years, ChatGPT and other generative artificial intelligence (AI) models have become mainstream, with many people having begun to wonder if this is something to fear, or something to embrace. Students and faculty at Cal Poly Humboldt have been testing the waters and, despite mixed reviews, it is apparent ChatGPT is here to stay. 

“I do think that there is a role for ChatGPT and gen AI models in the classroom. However, I think my recommendation for its use varies according to the level of the student,” said associate professor of computer science and software engineering program lead, Dr. Sherrene Bogle. “As higher educators, we have a responsibility to teach our students how to use the technology effectively.”

“ChatGPT is a tool applied in academia for diverse research like linguistics, psychology and computer science. It aids in hypothesis generation, language analysis and ethical exploration. Used in education and interdisciplinary studies, it’s showcased in conferences, with further advancements documented in recent academic publications,” according to ChatGPT itself.

Bogle first heard about ChatGPT last year and, although reserved at first, has since embraced it. Cal Poly Humboldt students studying computer science were offered an elective course this semester that focuses on AI, CS 480, taught by Bogle. One exercise she instructed in the lab was to use a selection of AI technology such as ChatGPT, Bard or Claude and find examples of where the programs hallucinate. 

 Noah Zerbe, a professor in the department of politics, defined what the term ‘hallucinate’ means for generative AI programs.

 “Sometimes [generative AI] will just make things up for you,” said Zerbe. “Often, one of the problems a lot of people have noted is that it makes up citations. So, it’ll say that a quote comes from this source, or something like that, and that source may not even exist. Or, it can go down a rabbit hole of factually incorrect information.” 

One mainstream example of ChatGPT ‘hallucinating’ is apparent in an incident where Colorado lawyer, Zachariah Crabill, was fired from Baker Law Group for using ChatGPT, and got caught when the program created fake scenarios in legal documents.

Another example of a generative AI model ‘hallucinating’ took place in the Cal Poly Humboldt AI class. According to Bogle, two students were able to convince an AI chatbot that they had lunch with Michael Jackson and Tupac. 

“They just said a few things, changed the prompt, and the technology apologized for saying that these people were dead and wanted to hear more about their conversation and was basically adding that to its database,” said Bogle.

As an advocate for programs like ChatGPT in academia, Zerbe encourages students to explore the program, but introduces generative AI models into the classroom with a warning.

“My approach to it is really centered on teaching students how to use it effectively,” said Zerbe. “It’s providing you feedback on your paper, trust it as much as you would trust feedback that you got from a friend down the hallway.”

Professors aren’t the only ones exploring this new territory with caution. Maddie Haus, a junior at Cal Poly Humboldt majoring in environmental studies, was suspicious at first. 

“I was stoked because it helped me so much, but I also found it scary because it seemed too good to be true,” said Haus. “I think it can be useful for achieving academic success. I do think there’s a thin line between doing the work yourself and just having ChatGPT do the work for you. Overall though, I think it’s a useful tool.”

Mikey Crispin, a Cal Poly Humboldt graduate who now works for the PBLC and the university as a scheduling and support analyst, has used ChatGPT on and off. 

“I didn’t really use AI for school because it didn’t explode until last November,” Crispin said. “By that time I was only taking the upper level computer science classes and could get all the info from professors.”

Now that he’s graduated, Crispin uses ChatGPT mainly for troubleshooting while writing software. 

“GPT doesn’t always know what you’re trying to do, but from a basic coding perspective, if you’re giving it info just to cover some gaps you might have, then it is a great assistant that can help you gain a better understanding of how something might work,” Crispin said. “I have no experience asking GPT for help for anything outside of a technological viewpoint. It did write a really good email for me once in the style of a Cat in the Hat book, so that was cool.”

All in all, ChatGPT, along with other generative AI programs, seem to be the future of education, but using AI as a crutch is inevitably harmful. 

 Bogle illustrates her perspective towards generative AI.

 “If you think about it, we don’t want our elementary school students and kindergarten students to be using a calculator, because we want them to learn to count, to carry their hundreds, tens, and ones, etc., to be able to do long division,” Bogle said. “But, after they have mastered being able to do their basic arithmetic with pencil and paper, we’re comfortable with them using the calculator. So, it’s similar with gen AI, which is why I said I think students need to master honing the particular skill first, before they become reliant on gen AI. That way, I think we have the best of both worlds.”