The Caribbean Examinations Council (CXC) will phase out traditional School-Based Assessments (SBA) in most non-practical CSEC and CAPE subjects, starting with CAPE in 2027 and CSEC a year later. The existing SBAs are being replaced by a supervised Paper 032, completed under exam conditions. AI expert Christopher Christie argues the move prepares students "for a world that no longer exists," opening one of the sharpest debates to confront Caribbean education in years.
CXC Registrar and CEO Dr Wayne Wesley said preserving the integrity of qualifications is the priority. "The integrity of our qualifications is not negotiable," he said, adding that while the council recognizes AI's potential, qualifications must continue to reflect a student's own knowledge and abilities.
Christie does not dispute that academic integrity matters. His criticism targets the speed of assessment evolution. "The problem is not AI," he said. "The problem is that our method of assessment has not evolved quickly enough to accommodate AI." Placing students in more controlled exam rooms, he wrote, will not prepare them for workplaces where they will be expected to use technology intelligently.
Rethinking assessment with AI transparency
Christie proposes what he calls an "AI-Transparent SBA." Students would openly declare any AI use, name the platform, and explain how it assisted their work. Instead of submitting only a polished final assignment, they would provide research notes, outlines, drafts, teacher feedback, and personal reflections. A short oral defence would follow, letting teachers probe the student's reasoning. Supervised classroom activities linked to the assignment would let teachers compare independent work with exam-condition performance.
According to Christie, a student who truly understands an assignment can readily explain the research process. A student who simply copied AI-generated material would struggle. He goes further, proposing that students be assessed on their ability to challenge AI. A history student might analyse an AI-generated account of the Morant Bay Rebellion for factual errors. An English student could critique and improve an AI-written essay. A principles of business candidate might evaluate an AI strategy for a Caribbean company, explaining why some recommendations fail locally. Even mathematics offers opportunities to find errors in AI-generated solutions.
"The student who can question that information, verify it and make a sound decision has demonstrated a higher level of ability than the student who simply memorises facts for an examination," he said.
Global shifts and the limits of detection tools
Christie's argument aligns with a growing international push to integrate AI literacy into education rather than exclude it. The global conversation around AI for Education has been shaped by UNESCO guidance encouraging systems to teach students about AI's opportunities, limitations, ethics, and risks. He also warns against overreliance on AI-detection software, pointing to a research paper he completed in 2016-years before generative AI was publicly available-that modern detection tools flagged as 98 percent AI-generated.
"The strongest protection is not detection software," Christie said. "It is assessment design."
A call to pilot before permanent change
Christie is urging CXC to pause the wholesale removal of traditional SBAs and instead pilot the AI-Transparent SBA in select subjects. Such a pilot, he said, should evaluate teacher workload, student performance, academic integrity, and the effectiveness of oral defences and supervised checkpoints. He wants educators, universities, employers, and technology specialists from across the Caribbean involved.
"We cannot prepare students for 2035 using an assessment philosophy designed for a world that no longer exists," Christie said.
Why this matters for educators
The debate signals a moment where educators can shape how assessment responds to AI, not simply react to it. Designing assignments that require students to declare, defend, and critique AI output demands a different skill set from teachers-one that combines content expertise with the ability to orchestrate oral defences and evaluate transparent research processes. Structured professional development, such as the AI Learning Path for Teachers, can support the transition. The choice between excluding AI or embedding it responsibly in coursework will define whether a qualification reflects a student's memory or their capacity to work with intelligent tools-and that decision sits squarely in the hands of the region's education professionals.
Your membership also unlocks:
