Professor Rose Luckin is Professor of Learner-Centred Design at the UCL Knowledge Lab in London with over three decades of experience working with AI in education. Luckin has become an advisor to policymakers and educators worldwide, consistently urging a focus on how humans actually learn, rather than what technology can easily do. A recent reflection in LinkedIn is particularly relevant for vocational education and training, challenging us to look past the marketing and consider the real nature of learning.

The conversation often starts with a familiar story, one Luckin recounts of a parent asking if they should invest in a premium AI tutoring service for their child. The tool explains things clearly and helps with homework, so what’s the problem? The problem, as Luckin powerfully argues, is not what these tools do, but what they don’t do. She highlights that today’s AI tutors, for all their sophistication, cater to a narrow slice of the human learning repertoire. They excel at exposition, rehearsal, and tutorial dialogue—explaining, drilling, and answering questions. While these are undoubtedly important activities, Luckin estimates they represent as little as 16 percent of the distinct ways humans learn.

This ‘16 percent’ figure suggests that in our rush to embrace AI-driven efficiency, we may be inadvertently promoting a dangerously incomplete model of learning. Luckin points out that we seem to have a collective blind spot about the mechanics of human learning, focusing only on the most easily automated components. We are optimising for what machines can measure and manage, and in doing so, risk marginalising the very processes that build deep, transferable, and lasting understanding.

For vocational education and training, this is more than an academic point; it is a direct challenge to the core practice of VET. VET is fundamentally about learning by doing, about building, collaborating, and solving authentic problems. It is about the messy, contextual, and often tacit knowledge that is developed in the workshop, the studio, or the workplace. When Luckin speaks of the importance of annotation for recasting knowledge, reflection for building metacognitive skills, and collaborative work for developing confidence, she is describing the bedrock of effective vocational pedagogy. These are not optional extras to be streamlined away; they are the essential activities through which a novice becomes a competent practitioner.

The danger, Luckin sidentifies, is that by treating the 16 percent as sufficient, we are not preparing learners for an AI-integrated future. Instead, we are preparing them to be dependent on AI. If a learner’s only experience of learning is being told information and drilled on facts, they will lack the skills to navigate complexity, collaborate with others, or adapt their knowledge to novel situations—the very skills that are becoming more valuable in an age of intelligent machines. The richness of an apprenticeship, the value of a work placement, the power of a project-based learning challenge—all lie in that other 84 percent.

This is not a call to reject AI tutors. Luckin is clear that they have their place, offering valuable support for practice, immediate feedback, and personalised pacing that can be a great benefit, particularly for accessibility. The problem is one of proportion and perspective. An AI tutor can be a valuable tool in the VET toolkit, perhaps for reinforcing foundational knowledge or providing targeted practice on a specific procedure. But it cannot and should not replace the rich, multi-faceted learning environment that is essential for developing true vocational competence. Our role as educators is to orchestrate that full, complex learning experience, using all the tools at our disposal, but never mistaking one of those tools for the entire orchestra.