- Opportunity, caution and the question of context
Artificial Intelligence (AI) has moved rapidly from abstract concept to everyday tool, and Sri Lanka’s education sector is no exception. From university classrooms experimenting with generative AI to school students using digital tools for revision, AI is beginning to shape how knowledge is accessed, explained, and assessed. Yet, as interest grows, so do questions around accuracy, ethics, equity, and relevance to local curricula.
Education policymakers, academics, and technologists largely agree on one point: AI is neither a cure-all nor an inherent threat. Its value depends on how it is designed, governed, and integrated into existing learning systems.
AI in higher education: what the research shows
Dr. M.R.K.N. Yatigammana, Senior Lecturer at the University of Kelaniya, has studied the use of AI tools in higher education institutions through a systematic literature review of 101 academic articles. Her research examined how AI tools affect student performance, engagement, and learning outcomes.
According to Dr. Yatigammana, the evidence suggests that AI, when used appropriately, can strengthen learning processes rather than replace them.”Across the literature, we observed that AI tools tend to support student engagement and improve access to learning resources,” she noted. “This has a positive association with academic performance, particularly when students use these tools as supplements to formal teaching.”
The study identified commonly used AI applications in higher education, with tools such as ChatGPT and other generative AI platforms appearing frequently. These tools were often associated with perceived benefits including improved understanding of complex concepts, time efficiency, and more personalised learning experiences.
However, the research also flagged clear limitations. Ethical concerns emerged as a recurring theme, particularly around academic integrity, over-reliance on AI-generated content, and uneven access to digital infrastructure. “The literature is clear that without proper guidelines, AI use can raise questions around plagiarism, assessment fairness, and data privacy,” Dr. Yatigammana explained.
Her findings point to the need for institutional frameworks rather than ad-hoc adoption. “Positive attitudes alone are not enough,” she added. “Investment in infrastructure, educator training, and policy development is necessary if AI tools are to support learning in a responsible manner.”
Dr. Yatigammana’s research ultimately suggests that Sri Lankan universities should focus on ethical governance and structured capacity-building, rather than banning AI outright or embracing it without safeguards.
From experimentation to structure
While universities grapple with policy and practice, AI adoption is also gaining momentum at the school level. Here, the challenges differ. Accuracy, syllabus alignment, and language accessibility take priority, particularly for students preparing for national examinations.
AI specialist Buddhika Jayanath argues that the conversation around AI in education should move away from extremes. “We often see AI framed either as a revolutionary solution or as something to fear,” he said. “In reality, it is a tool. The outcomes depend on how clearly we define its role.”
Jayanath stresses that generic AI systems, while powerful, are not automatically suitable for education systems with tightly structured curricula like Sri Lanka’s. “Most global AI tools are trained on broad internet data,” he explained. “That does not always translate well to local syllabi, marking schemes, or exam-focused learning.”
He also points to the importance of digital literacy among both students and educators. “If students do not understand how AI generates answers, they may treat outputs as unquestionable facts,” Jayanath said. “That can weaken critical thinking rather than support it.”
According to him, the role of educators remains central. “AI should assist explanation and revision, not replace teaching judgment or mentorship,” he added. “Teachers still provide context, values, and guidance that technology does not.”
The case for localised educational AI
One emerging response to these concerns is the development of AI tools built specifically around Sri Lanka’s education system. BrainUs, a local edtech startup, recently launched an AI-powered learning platform designed around national textbooks rather than open internet sources.
Founder and Lead Developer Tharushka Dinujaya Kumara says the motivation was rooted in trust and relevance rather than novelty. “Students don’t need an AI that knows everything,” he said. “They need an AI that knows exactly what is in their syllabus and nothing more.”
BrainUs uses a Retrieval-Augmented Generation (RAG) architecture, which means the system first searches an internal library of Ministry of Education textbooks and verified materials before generating an explanation. Each answer is linked to specific textbook sources and page numbers.
“This approach reduces the risk of confident but incorrect answers,” Kumara explained. “If the content does not exist in the curriculum, the system should not invent it.”
For Kumara, localisation is not simply a technical choice but an educational one. “When students prepare for O/Ls, accuracy matters more than creativity,” he said. “A single incorrect explanation can damage confidence.”
He also highlights language access as a major factor. “By grounding answers in Sinhala, Tamil, and English textbooks, we avoid awkward translations of English-first content,” he added.
Balancing innovation and oversight
Despite different approaches, experts broadly agree that AI tools must operate within clear boundaries. Dr. Yatigammana notes that ethical guidelines are as important as technological design. “Policies should clarify acceptable use, accountability, and transparency,” she said. “This applies to both global tools and local platforms.”
Jayanath echoed this view, emphasising that students should be taught how to use AI, not just given access to it. “Understanding limitations is part of digital literacy,” he said. “That includes knowing when AI is helpful and when human input is required.”
Kumara, meanwhile, argues that transparency is central to building confidence. “If parents, teachers, and students understand how an AI system works, they can judge its value more realistically,” he said.
Looking ahead
As Sri Lanka continues to explore AI in education, the path forward appears less about rapid adoption and more about thoughtful integration. Research suggests potential benefits in engagement and accessibility, while practical experience highlights the risks of misuse and overdependence.
What emerges from these perspectives is a shared emphasis on structure: clear policies, curriculum alignment, educator involvement, and ethical safeguards. AI, in this context, is neither an educational shortcut nor a replacement for teaching, but a support mechanism whose effectiveness depends on design choices and governance.
In the words of Dr. Yatigammana, “The question is not whether AI should be used in education, but how it can be used responsibly to support student learning.”
As the conversation evolves, Sri Lanka’s experience may offer a useful case study for other education systems facing similar questions, not about adopting AI quickly, but about adopting it carefully, with context and accountability at the centre.