Generative AI and academic misconduct: Helping students to navigate the grey

In their second blog on teaching AI to business undergraduates, Senior Lecturers at Liverpool Business School Gemma Dale CMBE and Mike Drummond CMBE focus on AI and academic conduct.

Earlier this year we shared on this blog details of our experimentation with bringing AI, particularly generative AI, into the classroom.  This academic year, we are continuing our work on teaching AI to business undergraduates.  This semester, we have been placing a particular focus on AI and academic conduct – a live issue for Higher Educations everywhere.

With some undergraduate assessments now indistinguishable from work produced by generative AI and media reports indicating that some universities are reverting to exams and handwritten work to verify authenticity, academic integrity in education is facing pressing challenges.

The current challenge

Generative AI has made cheating quick, easy and cheap.  Cheating is far from a novel issue in education.  Plagiarism and the use of essay mills are not uncommon, but the rise of tools like ChatGPT has turbo-charged the problem.  Most essay style assignments can now be completed to a reasonable standard in a matter of seconds, and typical processes for identify and addressing academic misconduct are often no longer fit for purpose in the generative AI world.  Academics are left playing catch-up as the situation evolves around them.

There is a need for educators to bring AI into the classroom in order to provide students with skills for the future. This must be balanced with academic integrity and maintaining standards. Tools will only become more powerful and sophisticated, and as such the risk of academic misconduct grows.  It is therefore critical - and urgent - for universities to establish clear guidelines that help students understand the boundaries between legitimate AI use and unethical practices.  This starts in the classroom.

Our approach

This semester, we sought to provide our business undergraduates with the resources, guidelines and practical examples necessary to help them use AI ethically, reinforcing the message that while AI can be a valuable tool, it cannot replace the value of genuine learning, effort, and original thought.

Of practical difficulty is the grey areas within this debate.  Whilst it is easy to convey that simply prompting generative AI with your assignment title and then undertaking a copy and paste of the answers is academic misconduct (and likely to result in a poor grade outcome) other examples are of generative AI use are not as straightforward.

For example, is it academic misconduct to ask a tool to help you improve your academic writing?  Is misconduct to ask a tool to generate ideas for your assignment or draft a script for a presentation?  Somewhat unsatisfactorily, the answer might be ‘sometimes’ or ‘it depends’. Where therefore, can that line be drawn between creative and beneficial use, and cheating? 

We started our session with a detailed discussion about academic misconduct, addressing the meaning of the term, key principles and potential outcomes for students.  We worked through examples of AI use in assessments, some of which were helpfully generated for us by ChatGPT, and facilitated conversations about whether they were or were not academic misconduct – or whether the answer fell into that grey area where ‘it depends on the circumstances’.  We combined these activities with wider discussions about quality of content.  Classroom activities were followed by a formative assessment with quiz style questions on academic misconduct to test learning. 

After we had completed our classroom activities, we ran a short survey.  To gauge how well students understood the acceptable use of AI in academic settings, we posed this question:

"How much do you agree with this statement: 'I understand how to use AI in my university work in a way that meets university requirements around acceptable academic conduct?'"

79% of students said that they agreed or strongly agreed with the statement.  The remainder were largely ambivalent – suggesting a need for ongoing guidance and real-life examples to solidify both understanding and confidence.  However, when we considered some of our qualitative responses, it was clear that some students remain concerned about accidental misuse or inadvertently breaking the rules when using AI to support them with university work. 

Future considerations

Student concerns about compliance are perhaps understandable given the ambiguous nature of the problem.  Definitive answers, and indeed academic misconduct policies, are difficult to construct in an AI world.  The only way to move out of grey areas is an outright ban – impracticable to enforce and a lost opportunity for our students who need to learn how to navigate the potential of Generative AI for their future careers. 

Of course, addressing academic misconduct is only one part of the bigger challenge.  The potential for cheating in academic work has always existed and will continue to, and procedures may need to be refreshed to ensure processes for investigating this continue to be robust.  At the same time as maintaining academic integrity, we also need to simultaneously move to new forms of assessment, reducing our reliance on testing what students know through traditional written assignments, replacing these with authentic and practical assessments. These assessments need to reflect what students need to know in the future of work.  Educators will also need to provide guidance to students on a case by case, assignment by assignment basis on how they may – or may not – use AI so they do not fall foul of accidental misconduct.  Without this, students will be lost in the grey area.