Dear Editor,

The Caribbean education system is confronting a difficult but necessary question: What exactly are we assessing when artificial intelligence (AI) can produce an essay, solve a problem, summarise a source, and imitate a student’s voice in seconds?

Recent developments at the Caribbean Examinations Council (CXC) and The University of the West Indies (The UWI) suggest that our assessment systems are beginning to respond.

CXC has announced major reforms to school-based assessment (SBA), driven in part by concerns about generative AI. From 2027, traditional SBAs will be phased out for non-practical subjects such as mathematics, English, Caribbean history, social studies, and principles of business, while SBAs will be retained for practical subjects. For the affected subjects, candidates will instead complete paper 032 under examination conditions, with topics provided approximately one month beforehand, additional completion time, and permission to bring reference notes.

At the tertiary level, The UWI has entered a partnership with the University of the West of Scotland to test IntegraGuard, a system that combines AI-assisted detection, investigation workflows, policy compliance, and academic-integrity reporting. These developments acknowledge an uncomfortable reality — the traditional written assignment is becoming increasingly difficult to use as straightforward evidence of individual thinking.

But replacing assignments and introducing detection systems will not, by themselves, solve the deeper problem. The real danger is not simply that students might ask a chatbot to write their essays, it is that educational systems might continue to equate writing with thinking. Writing is certainly an important manifestation of thinking, but it is not the only one. If AI can increasingly outsource the production of polished prose, educators need to create more opportunities for students to demonstrate the intellectual processes that precede and exceed writing.

This means developing critical thinking outside the traditional essay. One possibility is the use of oral defence and academic conversations. After completing a task, students could explain their reasoning, defend a conclusion, respond to challenges, and reconsider their position when presented with counterevidence. A chatbot can produce an answer; the student must demonstrate whether they understand and can defend it.

Debates and structured discussions offer another avenue. Students could be given competing claims and required to identify assumptions, evaluate evidence, and construct rebuttals. Importantly, assessment should reward the quality of reasoning, not merely the confidence of delivery.

Problem-based learning can similarly shift attention from products to processes. Students might be presented with an unfamiliar Caribbean social, economic, environmental, or educational problem and asked to investigate possible explanations, compare solutions, justify decisions, and respond to changing information.

There is also considerable value in observation, demonstration, and performance. In science, students can conduct and explain experiments. In business, they can analyse cases and defend decisions. In languages, they can participate in spontaneous conversations. In teacher education, they can respond to classroom scenarios. In the humanities, they can interrogate competing interpretations of a historical event.

Even more importantly, students should be taught to critique AI itself. Teachers can give students an AI-generated argument and ask them to identify unsupported claims, fabricated references, biases, missing perspectives, weak reasoning, and inappropriate generalisations. The chatbot then becomes an object of critical analysis rather than a substitute for intellectual work.

The Caribbean should, therefore, resist the temptation to frame the AI challenge simply as: How do we stop students from cheating? The more productive question is: How do we design learning so that genuine thinking becomes visible?

CXC’s reforms and The UWI’s IntegraGuard initiative represent responses to a rapidly changing educational environment. But academic integrity cannot ultimately be secured by surveillance or by removing writing from assessment. It depends on cultivating learners who can question, investigate, reason, communicate, defend, and revise their ideas.

In an age when machines can increasingly write for us, education must become even more intentional about teaching students how to think when nobody — or nothing — is thinking for them.

Oneil Madden

maddenoniel@yahoo.com