The rise of artificial intelligence in education poses an urgent threat to cognitive development if the technology is used indiscriminately.
A recent , led Dr Nataliya Kosmyna, found that ChatGPT users had the lowest brain engagement and performance compared with those using Google search or writing unaided, and they failed to integrate tasks into memory, risking the development of critical thinking and creativity.
While schools debate AI use in classrooms, the immediate priority is preserving academic integrity and fostering cognitive development.
Practical steps to reinforce cognitive development
The moment I first saw ChatGPT, it became clear to me that the world had changed. Recognising the importance of cognitive development, my department restricted AI use and restructured our departmental pedagogy in two key areas:
1. Coursework changes
Since September 2023, we have required that most of the research and drafting for A-level coursework be completed in class. This policy is particularly important given the rise of tools such as HyperWrite History AI and WriteAI.
Implementation
Previously, students focused on researching key texts during lessons and did most of their draft writing at home.
We have now transitioned to a more structured system, whereby lessons are split between researching key texts and drafting coursework.
Progress tracking
Using Google Docs, we monitor each step of students’ work, from initial ideas to final drafts.
This real-time oversight ensures transparency and helps to identify any inconsistencies in tone or sophistication that may be generated by the use of AI.
Checking for authenticity
As part of the coursework process, we are allowed to provide one draft with general feedback. I have added an extra step to assess students’ understanding of their work.
If a student submits complex arguments that they are unable to explain, it raises questions about authenticity. Fortunately, we have not encountered any issues so far.
Impact
It is heartening that much has remained the same. Reassuringly, students are still finding the work as challenging as pre-AI cohorts and displaying similar behaviours, such as requesting extensions and moaning about having to do coursework.
However, one welcome benefit of drafting coursework during lessons has been a notable reduction in student procrastination.
Challenges
Despite these measures, a student determined to use AI can still find ways around the checks. While coursework reform is inevitable, we remain committed to upholding intellectual and ethical integrity within the current system.
2. Essay writing changes
Recent by Professor Audrey van der Meer and colleagues shows that handwriting improves memory and recall by encouraging more deliberate thought. At the same time, the shift to laptops during the Covid crisis has caused a noticeable decline in handwriting skills.
To address this, we introduced a policy requiring handwritten classwork and homework for Year 7 to Year 13 students, with laptops used only sparingly.
This negates the temptation to copy and paste AI-generated answers - a concern that weighs heavily on us, especially for examination groups.
Impact
The reduction of laptop use has meant fewer distractions among students, and handwritten work across the key stages has broadly improved, although we accept that it was easier to collate and mark work on Microsoft Teams.
Challenges
We work closely with the individual needs department to determine whether students should use laptops in exams, ensuring that our approach is flexible and responsive. To support this, we make decisions early, giving students plenty of time to build their typing speed and confidence ahead of summative assessments.
Ethical and pedagogical discussions
We have engaged with key stage 3 and 4 students about the ethical implications of AI and the risks of blindly trusting it. AI tools remain prone to “hallucinations”, often producing inaccurate historical or scientific information.
For example, we have discovered cases where students using AI to generate pictures for homework have produced bizarre errors, such as Spanish Armada galleons at the Battle of Jutland and elephants at the Battle of the Somme.
Discussing these AI-generated images with historical inaccuracies has helped us to engage students in critical thinking and the reality that AI is not foolproof.
Younger students
We are less concerned about AI with younger students because most of our tasks for them involve using textbooks and classroom materials, with the majority of written work completed during lessons.
However, we have instilled a sense of caution in our students about using AI, framing cognitive offloading in terms that they can easily understand; namely that AI is not a short cut for their own ideas.
What does the future hold?
As AI becomes an unavoidable part of education, schools must intentionally integrate it into their curricula, ensuring that it supports learning without replacing critical-thinking skills.
However, until we can be sure that AI is supporting rather than supplanting that process, restraint in adopting it in our classrooms is not resistance but a duty of care.
David Tuck is head of history and politics at Harrow Hong Kong
For essential weekly intelligence on the international schools sector, sign up for the
