ChatGPT: cheating engine or useful resource?

Just over two months since the launch of the new artificial intelligence (AI) tool ChatGPT, Jamaican educators are grappling with the question of ensuring academic integrity even as some point to the merits of the chatbot.

ChatGPT, which was developed by Silicon Valley start-up OpenAI and launched in November 2022, has powerful capabilities allowing it to write sophisticated human-like essays, pass MBA and Bar exams, complete assignments and generate creative works, among other things, forcing some school districts in the United States to temporarily ban its use on school computers.

Several Canadian universities are crafting policies on ChatGPT and lecturers in the United Kingdom have been urged to review assessments.

Professor Colin Gyles, acting president of the University of Technology (UTech), Jamaica, underscored that students have always been able to access external assistance before the advent of ChatGPT or any other AI software.

“The process now appears easier and brings into question whether we are capable of producing knowledge-based workers,” he reasoned.

Gyles said that ChatGPT and other AI tools are here to stay and educators must adapt to the technology and familiarise themselves with the tool’s capabilities, deficiencies and develop ways to use it in instruction.

But he reasoned that assessments will have to be restructured to ensure that learning is being measured and not just standard correct answers.

“In the short term, we will control AI use on campus networks, utilise detection software, revert to handwritten monitored assessments or examinations, apply multiple assessments and review our academic integrity policy regarding examinations and plagiarism. Curricula will have to be revised and this revision will include the course objectives and assessment methods,” Gyles told The Sunday Gleaner last week.

In an article posted on its website on January 31, OpenAI said it had launched a classifier trained to distinguish between AI-written and human-written text.

However, OpenAI cautioned that its classifier is not fully reliable.

“In our evaluations on a ‘challenge set’ of English texts, our classifier correctly identifies 26% of AI-written text (true positives) as ‘likely AI-written’, while incorrectly labelling human-written text as AI-written 9% of the time (false positives),” a section of the article read.

UTech’s acting president also told The Sunday Gleaner that an AI literacy education and training programme for lecturers will have to commence with urgency, so that academic staff can leverage AI tools for dynamic learning and instruction and institute proper safeguards to avert any abuse of AI tools.

Further, Gyles said that UTech is currently in dialogue with the deans of colleges and faculties to engage faculty members in testing the software against the assessment instruments that they currently use to see the extent to which they might need to make adjustments in their assessment methods and approaches.

Meanwhile, University of the West Indies, Mona Campus Registrar Dr Donovan Stanberry underscored that academia has concerns about ChatGPT.

“It is an emerging phenomenon, so I don’t think we have a settled position. On the surface, yes, everybody would have concerns. I think it will challenge our lecturers to be more creative in terms of how they examine students, not just for recall but for critical thinking, but, I am sure in due course, we will have a settled position,” Stanberry told The Sunday Gleaner.

Darien Henry, principal of Montego Bay Community College, posited that the intensification of media technologies has both positive and negative repercussions and it was always contemplated that the impact on education would be far-reaching, but it is happening at a far more rapid pace.

“ChatGPT had over a million users in the first three months of its release in November last year. Caktus is another AI-based technology which is even more intuitive and human-like and cuts across many disciplines. Students have always been finding ingenious web-based ways of completing assignments and COVID-19-induced virtual schooling intensified that. School leaders have to be aware of this and be cognisant about the impact of pedagogy,” he said, adding that last Monday he did a presentation to his executive leadership team about the implications of AI technologies on higher education and possible ways they could respond.

Henry detailed that some high schools, colleges and universities in the United States, Australia, and Europe are on the verge of mainstreaming AI technologies that enrich teaching and learning, enrolment, academic advising, student retention and overall student engagement.

“School leaders must now contend with demographic shifts in the learning attitudes of students who are younger, agile and technologically nimble. Therefore, school improvement planning, strategic planning, and policy changes must seriously consider the implication on the teaching and learning processes and the outcomes as well,” the educator explained.

Henry underscored that the integration of Al-based technologies has certainly shifted the dynamics for assessing students, determining the pace of students’ critical thinking and how they solve problems.

He said this means that students have to be prepared to interact with these technologies responsibly and in a way that excites and engages them in the learning process.

Further, he said this also means that educators will have to push the importance of academic honesty when preparing and evaluating lessons.

judana.murphy@gleanerjm.com