Fears of cheating over chatbots were exaggerated, new research suggests

When high school and college students started trying out a new AI chatbot called ChatGPT to complete writing assignments last December, fears of mass cheating spread across the United States.

To prevent bot-enabled plagiarism, some large public school districts – including those in Los Angeles, Seattle And New York City — ChatGPT quickly blocked on school-issued laptops and school WiFi.

But the alarm may have been overblown — at least in high schools.

According to new research from Stanford University, the popularization of AI chatbots has not increased the overall rate of cheating in schools. Surveys this year of more than 40 U.S. high schools show that about 60 to 70 percent of students have recently engaged in cheating — about the same rate as in previous years, says Stanford Education researchers said.

“There was a panic that these AI models would enable a completely new way of doing something that could be perceived as cheating,” he said. Denise Popea senior lecturer at the Stanford Graduate School of Education who has been surveying high school students for more than a decade a non-profit education organization she is a co-founder. But “we just don’t see the change in the data.”

ChatGPT, developed by OpenAI in San Francisco, began capturing the public imagination late last year with its ability to craft human-sounding essays and emails. Almost immediately, classroom technology boosters began promising that AI tools like ChatGPT would revolutionize education. And critics began warning that such tools — which liberally fabricate things — would enable widespread cheating and amplify misinformation in schools.

Now the Stanford researchalong with a recent report from the Pew Research Center, challenge the idea that AI chatbots are upending public schools.

Many teens know little about ChatGPT, Pew found. And most say they’ve never used it for schoolwork.

Of course, these trends may change as more high school students become familiar with AI tools.

Many teens have never heard of ChatGPT

This fall, the Pew Research Center surveyed more than 1,400 U.S. teens ages 13 to 17 about their knowledge, use, and views of ChatGPT. The results may seem counterintuitive given the plethora of panicky headlines last spring.

Nearly a third of teens said they had heard “nothing at all” about the chatbot the Pew surveyconducted from September 26 to October 23, 2023. Another 44 percent said they had heard “a little” about it.

Only 23 percent said they had heard a lot about ChatGPT. (The Pew survey didn’t ask teens about other AI chatbots like Google’s Bard or OpenAI’s GPT-4).

Responses varied by race and household income. About 72 percent of white teens said they had heard of the chatbot, compared to about 56 percent of black teens, Pew said.

About 75 percent of teens in households with an annual income of $75,000 or more said they had heard of ChatGPT, Pew found, compared to just 41 percent of teens in households with an annual income of less than $30,000.

Pew also asked teens if they had ever used ChatGPT to help with their schoolwork. Only a small minority – 13 percent – ​​said so.

The results of the Pew survey suggest that, at least for now, ChatGPT has not become the disruptive phenomenon in schools that advocates and critics predict. Among teens who said they had heard about the chatbot, the vast majority (81 percent) said they had not used it to help with their schoolwork.

“Most teens have some awareness of ChatGPT,” said Jeffrey Gottfried, associate director of research at Pew. “But this doesn’t even account for the majority of teens who are already integrating it into their schoolwork.”

Cheating rates haven’t changed much

Cheating has long been rife in schools. From surveys among more than 70,000 secondary school students between 2002 and 2015, 64 percent said they had cheated on a test. And 58 percent said they had plagiarized.

Since ChatGPT’s introduction in 2022, the overall frequency of high school students reporting that they have recently committed fraud has not increased, the Stanford researchers said.

The new research does not shed light on how often students use chatbots as cheating bots. The Stanford and Pew researchers did not survey students about their use of AI tools.

This year, the Stanford researchers added survey questions that specifically asked high school students about their use of AI chatbots. This fall, 12 to 28 percent of students at four high schools on the East Coast and West Coast said they had used an AI tool or digital device – such as ChatGPT or a smartphone – in the past month as an unauthorized aid during a school test, assignment or homework.

Of high school students who said they had used an AI chatbot, about 55 to 77 percent said they had used it to generate an idea for a paper, project or assignment; about 19 to 49 percent said they had used it to edit or supplement part of an article; and about 9 to 16 percent said they had used it to write an entire paper or other assignment, the Stanford researchers found.

The findings could help shift discussions about chatbots in schools to focus less on fear of cheating and more on helping students learn to understand, use and think critically about new AI tools, the researchers said.

“There are other ways to think about AI – not simply as this uncontrollable temptation that undermines everything,” he says Victor R Lee, an associate professor at the Stanford Graduate School of Education who researches AI learning experiences and, along with Dr. Pope led the recent fraud investigation. “There is so much more that could and should be talked about in schools.”

‘Not acceptable’ for essay writing

While schools are still developing acceptable usage rules for the AI ​​tools, students are developing nuanced views about using ChatGPT for schoolwork.

Only 20 percent of teens ages 13 to 17 said they found it acceptable for students to use ChatGPT to write essays, Pew found. But almost 70 percent said it was acceptable for students to use the AI ​​chatbot to research new topics.

This doesn’t mean that students won’t try to pass chatbot-generated texts as their own schoolwork.

Christine Meade, an Advanced Placement history teacher at a high school in Vallejo, California, said cheating with chatbots was widespread among 12th graders last spring. She even caught a few using the AI ​​chatbots on their smartwatches during school tests.

But this year, after telling her students they could use ChatGPT and Bard for certain research projects, the situation “completely changed,” she said.

“I had some students in my AP history class use chatbots to generate a list of events that happened just after the Civil War, in the 1880s,” Ms. Meade said. “It was pretty accurate – except for what happened in the 1980s during the Reagan administration.”