A recent article published by The Times has reported that the number of Irish college students that cheated on assessments using AI has tripled in the last two years. The Times’ data come from institutions like DCU, RCSI, and the University of Galway. Notable absences, though, are UCD (with no data available at all from the past three years) and Trinity College, whose last data collection on AI misuse goes back to 2023–24. A spokesperson for Trinity Communication has confirmed that indeed Trinity does not hold data on AI-led misconduct.
The same article reports that, rather than aiming to stop the use of AI, Trinity is instead trying to learn how to live with it — and maybe even how to exploit it. The use of AI is not immediately treated as misconduct by most schools and departments: as a spokesperson reported, most Schools reflect the College Statement on Artificial Intelligence and Generative AI in Teaching, Learning, Assessment and Research, which states that Trinity aims to support staff and students to become “AI literate and fluent”: in summary, to be able to recognise and use AI’s benefits while managing its risks.
For example, the School of Mathematics 2025–2026 Handbook states that “the use of GenAI is permitted unless otherwise stated”, adding that, when GenAI is used, it has to be acknowledged and cited properly; the PPES 2025–2026 Handbook recites the same. In summary, in these schools, GenAI is permitted as long as the user states clearly when they do so. By itself, this might not seem wrong: for scientific courses and subjects, refusing a new progressive development upfront and as a whole can be more harmful than not. Therefore, it is correct to take a new discovery or invention into consideration and assess its good sides, as well as its faults. But with the rate of cheating through GenAI going up across the board, a few departments have had to address this matter in their handbooks.
In the past few years, the School of English has had to introduce in-person exams in order to combat the essays written by generative AI. But in-person exams risk neutralising a huge part of an English degree assessment, which is the ability to research secondary sources, analyse them thoroughly, and write new, original essays based on them. In-person exams leave little space for this, as their preparation requires memorisation of primary sources and the ability to think and write quickly, instead of thoughtfully and while developing a personal style.
Moreover, because an English degree requires students to summarise, produce, and analyse texts as their basic learning outcomes, the School of English (and similarly does the School of Languages, Literatures and Cultural Studies) states in its Handbook, exactly contrary to the other Handbooks abovementioned, that GenAI is not permitted unless in very specific circumstances. They specify that “Generative AI tools may enable advances in other disciplines, but they are of very limited use in the meaningful study of English Literature.” The handbook then explains that “written assessments in English require original critical thought and analysis — not the unreliable synthesis of information from unknown sources” (School of English 2025/2026 Handbook).
In summary, the humanities schools have had to start changing their structures to preserve fairness in both teaching and learning. So while learning to use and continue to develop Generative AI can be useful for most lab-based or scientific subjects, it also creates an extremely harmful instrument available for all to use freely. The more GenAI is studied and therefore developed, the more it becomes harmful to other branches of study; ignoring this means to implicitly favour scientific subjects over humanities as they are considered less important, as if human development was only made through science and technology.
AI literacy is fundamental to scientific courses; but the use and development of it is fundamentally harmful to humanities (and all those middle-ground courses like Psychology or Law), threatening them ontologically. Giving a general go-ahead on AI is a risk that is not worth taking, when it could easily be curbed by confining the use of AI solely to specific courses and schools that need it.
Given the speed at which AI development is going, AI use and related academic misconduct seeps through assignments and it becomes more and more difficult to catch. It is complicated to keep track of AI-led misconduct, especially if AI is generally permitted in assessments. So instead, its use becomes accepted and inevitable, appealing even to those courses that will only be negatively affected by it in both the short or long term. So, how far does the scientific interest in development have to go, before we realise that it went too far?
