Artificially intelligent chatbots are pushing schools to keep up with modern cheating 

• Shutterstock

What was the best excuse you used on a teacher for turning in homework late? Justified or not, most students have had something come up—or come up with something—at least once in the face of an academic deadline.

Maybe you said you were too sick to work, or—if you were truly bad—that your grandma died. Perhaps it's just part of the human experience—there's no judgment here.

And whatever the justification a student concocts, the days of making up excuses to miss class-work deadlines may no longer be necessary. Why lie when you could have a robot do your homework for you?

Recently developed, artificial intelligence-powered chatbots—such as OpenAI's ChatGPT—can write poems, essays and presentations for procrastinating students in a matter of minutes. All it takes is a simple prompt.

These bots are a powerful new resource at the fingertips of every student who has an internet connection. However, the chatbot tools can also be used to lightly dress up plagiarism in the classroom—a growing concern for colleges and school districts across the nation.

In New York and Seattle, ChatGPT has been banned on school Wi-Fi networks and devices, though students can easily find workarounds to access it.

Other districts are re-imagining their academic policies and purchasing anti-plagiarism software to combat the use of chatbots. So far, it's unclear whether, or how, Utah educators will follow suit.

First, what are chatbots and how do they work? Software labs like OpenAI and Google have created bots to operate like a personalized search engine, only instead of directing you to Wikipedia or .edu pages, the bots generate original responses themselves.

Instead of Googling "Battle of Waterloo," a user can ask ChatGPT or LaMDA, "What happened at the Battle of Waterloo?" and it will write an answer. The bot can take things further by turning those responses into essays, limericks or stand-up comedy routines (though jokes about Napoleon's failure are probably past their prime).

Users can ask the bot to regenerate the response to use more humor or to beef up the academic vocabulary. And the responses are eerily nuanced and articulate, enough to raise alarm bells for Antony Aumann, a world religions professor at Northern Michigan University.

In interviews with national media outlets, Aumann describes incidents like that of a student's essay that was well-written and organized, but seemed off somehow. When confronted, the student admitted to using ChatGPT and ever since then, the professor has revamped how he delivers coursework.

"All of a sudden, you have someone who does not demonstrate the ability to think or write at that level, writing something that follows all the requirements perfectly with sophisticated grammar and complicated thoughts that are directly related to the prompt for the essay," Aumann told Business Insider.

He told the New York Times that his students now write first drafts of essays in the classroom using browsers that monitor computer activity, and students have to explain each revision of that draft in future submissions. Aumann went on to say that he may have to do away with essay assignments entirely in subsequent semesters.

ChatGPT, in particular, has caused quite a hubbub among academic institutions in the United States since its November 2022 release, though OpenAI CEO Sam Altman has downplayed the power of the bot, telling The New York Times in an interview that "ChatGPT is a horrible product."

"ChatGPT is incredibly limited, but good enough at some things to create a misleading impression of greatness," Altman later went on to say in a December tweet. "It's a mistake to be relying on it for anything of import right now. It's a preview of progress; we have lots of work to do on robustness and truthfulness."

Altman is right in that bots like ChatGPT come with plenty of limitations. ChatGPT has a character limit of about 500 words and has a server capacity that locks out users if it's too busy. And it is only familiar with events that occurred before 2021 and tends to inaccurately quote sources.

Still, Microsoft invested $10 billion in OpenAI to integrate artificial intelligence into its search engine, Bing, along with other projects for its lab. Indeed, the so-called generative language models are in their infancy, and likely to grow in popularity. That's concerning for college professors and K-12 teachers alike, but it could also have potential as an instructional tool.

Aumann said one way he plans to weave ChatGPT into lessons is by asking his students to evaluate its responses.

"What's happening in class is no longer going to be, 'Here are some questions—let's talk about it between us human beings,'" Aumann said. Instead, he told the Times, "it's like, 'What also does this alien robot think?'"

With bots like ChatGPT gaining popularity, how will Utah schools respond?

School districts in Utah are eyeing artificial-intelligence language models like ChatGPT with varying levels of speculation.

Teachers in Granite School District have not yet reported specific instances where chatbots were used on assignments, but they've begun evaluating student work with extra attention on signs of AI-generated plagiarism, said Matthew Sampson, a spokesman for the district, via email.

Using AI chatbots to replace writing is a violation of Granite's student code of conduct, Sampson said. And while Granite makes plagiarism detection software available for teachers, they aren't yet equipped to handle the sophistication of next-generation chatbots.

"Since ChatGPT can be utilized to generate a variety of writing, we are hopeful that we will be able to provide our teachers with technology and training soon that will help curb this form of AI-generated writing from being submitted as original work," Sampson said.

Sampson said the district's software, Turnitin, is also working on upgrades to detect artificial writing going forward.

Cheating resources are nothing new, Sampson added. In the future, he said the district will follow the plagiarism and cheating policies currently in place. And he noted that teachers have used methods like classroom presentations to evaluate student understanding beyond writing long before the advent of generative chatbots.

At this point, no discussion has taken place to update academic integrity policies and specifically ban chatbots like ChatGPT in Granite. The district will instead evaluate the implications of the new tech on classroom instruction.

But some teachers across the U.S. have also started to experiment with incorporating AI tools into their coursework, Sampson said, and Granite could consider asking students to evaluate and improve on AI-generated responses as part of a presentation.

"We understand that technology will continue to adapt and change and there may be a possibility of adapting our instructional practices to use AI technology as an instructional tool," Sampson said. "However, we want to fully understand the ramifications of using AI technology before introducing it in our classrooms."

But the use of chatbots raised enough concern in Davis School District—the state's second-most populated district—that, on Jan. 24, assistant superintendents Logan Toone and John Zurbuchen gave a presentation to the school board during a study session on the bots' capabilities and how to maintain student creativity.

And in Salt Lake City School District, administrators haven't yet discussed the use of chatbots in any official capacity, though spokesperson Yándary Chatwin indicated that some district leaders do not think that a policy change would be in order.

The district already has standards for academic honesty and any changes would likely be related to delivery of instruction, Chatwin said. Officials haven't talked about how.

Utah school districts are largely betting that the early phase of the battle against AI will be waged more in individual classrooms and less in broad board policies. So as the use of bots does increase, what can administrators and teachers do to prevent plagiarism?

As Granite spokesman Sampson mentioned, districts can wait for the anti-plagiarism software they already use to be updated to catch AI-generated content. OpenAI is also developing software designed to identify text written by AI, according to The New York Times.

More than 6,000 teachers from Harvard University, Yale University, the University of Rhode Island and others have also signed up to use GPTZero, a program that promises to quickly detect A.I.-generated text, Edward Tian, its creator and a senior at Princeton University, told The New York Times.

Postsecondary schools have been frenzied since the release of ChatGPT, with universities establishing task forces and hosting discussions on how to respond to the tool. Professors are phasing out the once dominant method of sending students home with open-book assignments, like they had during the pandemic.

Instead, students are doing more in-class assignments, handwritten papers, group work and oral exams, according to national reporting.

Gone are prompts like "write five pages about this or that." Rather, some professors are crafting questions that they hope will be too clever for chatbots, or are asking students to write about their own lives and current events.

Others are raising their standards for grading essays. A thesis and a few supporting paragraphs isn't enough for an A anymore, especially when a robot can do it easily.

Some universities and school districts outside of Utah are drafting revisions to their academic integrity policies so their plagiarism definitions include generative A.I.

No matter what Utah's approach is to stopping cheating using AI-generated chatbots, the bots are here to stay. Students are sharing on forums like Reddit that they have submitted assignments written and solved by ChatGPT. On TikTok, the hashtag #chatgpt has more than 578 million views, with people sharing videos of the tool writing papers and solving coding problems.

One video identified by The New York Times shows a student copying a multiple-choice exam and pasting it into the tool with the caption saying: "I don't know about y'all but ima just have Chat GPT take my finals. Have fun studying."

And there are bots that can do much more than respond to simple writing prompts. Mindgrasp is a software that can summarize and create extensive notes from webpages, videos or PDF uploads. Other bots, like QuillBot, are designed to dress up AI-generated text from models like GPT to sneak around plagiarism software.

It feels too futuristic to be true, but students are using chatbots now that will only get smarter going forward. Hopefully, kids will use the bots to generate better excuses for missing their assignment deadlines, rather than generating the assignments themselves.