THE current educational debate over the impact of generative artificial intelligence learning is because this technology is different from earlier AIs. The latter helped manage knowledge, summarise data and analyse and synthesise bulk information. GAI, with its advanced algorithmic architecture, generates new content in text, images and videos. It is easier, faster and accessible, which raises concerns about knowledge automation as GAI can produce information, solve problems and prompt learners to automate assignments. Dependence on automation harms cognitive capabilities.

Research affirms the correlation between GAI and cognitive erosion, weakening neural signals, reducing critical thinking scores and impairing independent recall. In the short term, GAI can enhance speed and efficiency. In the long run, it can diminish learners’ intent to learn, think critically, discover, solve real problems and make decisions. The education systems most prone to cognitive decline are those that promote cramming and learning by rote, and have poor teacher support and foundational learning.

In an exam-cramming culture, GAI becomes a convenient tool to produce summaries, answers and model essays. Students spend less time in idea generation; increased prompting and information production create an illusion of competence. The approach often weakens retention and comprehension. This knowledge ‘procurement’ is not conducive to real problem-solving as it was never deeply encoded.

Memory-based education negates deeper learning — elaboration, connecting ideas, explaining causes, comparing alternatives and applying knowledge in new contexts. GAI use makes students memorise what machines produce, transitioning from rote learning to rote dependence. They may reproduce acceptable text but not know why the answer is correct, what assumptions it rests on, or how to challenge it. Their critical thinking, which depends on judging and evaluating claims, stands compromised.

Human agency must be prioritised over content coverage.

Moreover, weak teacher support or poor collaborative learning between teachers and students has ample space for readily available help. GAI is appreciative, non-judgemental, non-threatening and instant. Additionally, poor foundational learning results in students struggling to read, write, comprehend and reason. In 2024, the World Bank noted high learning poverty. This means about 80 per cent of children in the later stages of primary school are not proficient in reading.

Weak reading abilities then result in increased GAI usage to mask literacy issues. Later on, this leads to a false sense of proficiency that ultimately hampers long-term educational development.

All these fault lines make GAI an instrument more devastating for cognitive erosion than any other learning tool. As a solution, the exam system needs to be restructured so that board exams are included as the more analytical method. This will require rethinking standardised exams and making them school-based.

Overloaded and content-heavy curricula and textbooks will always make GAI a first option as it makes learning far easier. Policies should, therefore, reduce heavy syllabi and move towards competence-centric curricula that explicitly reward explanation, argument, comparison and problem-solving.

A shift that prioritises human agency over content coverage is vital. This should come with new skills such as AI literacy, data literacy and self-regulated learning.

A system riddled with an unprofessional teaching force cannot benefit from GAI. That means the ri­­ght policy response is not to simply introduce AI but rather to professionalise teacher support for AI-mediated le­­arning. Outdated and old-fashioned instructional prof­e­ssional developm­ent has become a bu­­rden. It has to go beyond the implementation of a fixed syllabus. Teachers must act as informed designers of learning experiences that are tailored to cater to the diverse needs of students and incorporate effective strategies for GAI integration in the classroom.

We must ensure early-grade reading and numerical proficiency are top prioritie. We must restrict high-stakes GAI use in the lower grades until foundational skills are secure. GAI tools can be used moderately in schools to support local languages and age-appropriate reading levels, and to improve AI literacy with AI pedagogical and content knowledge.

The real threat is not GAI, but the education system’s flaws that compel students to rely on it.

The solution is not to restrict GAI but transform education through strengthened foundational skills, redesigned assessment and empowered and updated teachers who are curriculum developers so that GAI use encourages critical and independent learning.

The writer is an educationist.

Published in Dawn, July 7th, 2026