June 29, 2023

Ways Students Use AI To Cheat and How To Stop It

The rapid rise of AI-driven tools like ChatGPT places educators in a difficult position. Artificial intelligence can enhance the educational experience but can also be used for cheating. Some schools have already taken a stand by banning the technology. But is there a healthy way to unleash AI in the classroom without undermining the goals of education?

When schools nationwide began banning ChatGPT following its November 2022 release, they were clear about their fears. “Like all school districts, Seattle Public Schools does not allow cheating and requires original thought and work from students,” said Tim Robinson, a spokesperson for Seattle Public Schools, when asked by GeekWire why the district had blocked access to the chatbot.

How exactly did educators believe the generative AI chatbot and other AI-driven tools would empower students to cheat? Here are a few examples.

By Joseph Wilson, Co-Founder — Studicata

How AI can be used for cheating

The generative AI that drives ChatGPT makes it an excellent writing tool. For those suffering from writer’s block, it can provide opening paragraphs for virtually any writing assignment and clean up writing to improve grammar and style.

If that was the extent of AI’s potential, educators might not be so concerned, but ChatGPT can also spit out a complete essay on virtually any topic in seconds. Students using it to that end are not contributing to the original thought that educators want to see.

Teachers who assign take-home exams or who fail to proctor exams stand the risk of having students use AI-driven programs to provide answers they can then cut and paste into their submissions. In a best-case scenario, students would use AI to assist with language construction, turning to it to improve organization and smooth out clunky responses. The fear, however, is that AI becomes the “respondent,” meaning the student’s level of learning has not been effectively assessed.

More sophisticated students could use AI tools to enhance their workflows. They could query AI programs for an outline for a paper, request recommendations for sources, or even ask for positions or approaches they can take to enhance their papers and, consequently, their learning.

This begs the question: Is it cheating to use AI to help frame responses that students flesh out or to help clean up writing once it has been done? Most students are still waiting for a definitive answer. Where schools have not outrightly forbidden AI assistance in assigned coursework, students are operating in a “gray area” with academic guidelines that fail to address the realities of life in the age of AI.

How educators can stop AI abuses

Teachers can turn to AI detection tools to keep AI abuse out of the classroom. Many plagiarism detection tools that educators are already using — such as Turnitin and CrossPlag — have announced they are working on enhancements to their platforms that will detect AI-enhanced writing.

Teachers can also use ChatGPT-specific detection tools like ZeroGPT or OpenAI’s AI text classifier to reveal assignments that have benefitted from the tool. OpenAI has also indicated that they will soon include watermarks in the text created by ChatGPT, which could aid in detecting AI-generated submissions.

It is important to note that AI detection tools are in the early phases of development and are not foolproof. Testing has shown that they can fail to identify text that is AI-generated, and they have also been shown to sometimes falsely flag text written by a human.

Alternatively, educators can also creatively adjust how they structure their tests and assignments to limit the usefulness of AI. This can include conducting oral exams in the classroom that gauge understanding at the moment rather than assigning work completed in environments where AI tools can be used.

Teachers can also limit the usefulness of AI by focusing assignments on “breaking news” and other current topics that do not pre-date the release of the most recent versions of AI tools, as generative AI tools provide answers based on the training they have received. ChatGPT 3.5, for example, was trained on data through 2021, which limits its ability to respond to more recent events. Focusing on niche topics or obscure readings is another way to keep students from cheating by using AI tools.

How the education model must evolve

The rapid rise of AI-driven tools has been a disruptive force in the education industry, and as educators wrestle with the best way to respond, they shouldn’t think that AI will fade away. The challenge that today’s teachers face is how to prevent cheating and prepare their students for a future in which AI tools will be a part of their daily lives.

A proper balance involves seeing AI programs as tools that can assist students with their coursework but not replace the effort they must put into their studies. It will be up to teachers and schools to define how much AI is encouraged and then help their students navigate the line between cheating and assisting.

— Joseph Wilson is the Co-Founder of Studicata. He can be reached at joe@studicata.com.

About the author 

Elle Gellrich

{"email":"Email address invalid","url":"Website address invalid","required":"Required field missing"}