Artificial Intelligence has entered education, training and assessment faster than most institutions could prepare. The first reaction was predictable: panic about cheating, plagiarism and whether learners are “using AI”. But this is the wrong starting point. The real question is no longer simply whether AI was used. The deeper question is whether the learner, student, employee or professional still genuinely understands, owns and can defend the work produced with AI support.
This is where the idea of Cognitive Authenticity becomes important.
Cognitive Authenticity extent is the degree to which a person’s final AI-assisted work still reflects their own thinking, judgement, reasoning, understanding and ethical agency. In simple terms: can the person explain what was produced? Can they defend the argument? Can they identify weaknesses? Can they show what they added, changed, rejected or improved? If not, we may not be looking at assisted learning. We may be looking at outsourced thinking.
This distinction matters deeply for the post-school education and training sector. APPETD members work in a system where competence, workplace readiness and credible assessment are central. If used well, AI can support learning, improve access, strengthen writing, help learners organise ideas and provide feedback. But if used badly, it can create a dangerous illusion of competence. A learner may submit polished work with very little actual understanding behind it. That is not only an academic integrity problem. It is a quality assurance problem.
The old question was: “Did the learner write this?” The new question should be: “Can the learner think this through?”
This shift has significant implications for providers, assessors, moderators and workplace learning partners. AI detection tools alone will not solve the problem. They are often unreliable and focus too narrowly on the text’s origin. We need to focus instead on the authenticity of the thinking. A learner should be able to demonstrate the process behind the product: the original idea, the prompts used, the sources checked, the decisions made, the corrections applied and the learning gained.
A practical Cognitive Authenticity approach would assess six criteria. First, intentionality: did the learner begin with a clear purpose of their own? Second, conceptual ownership: can they explain the key ideas in their own words? Third, reasoning transparency: are the steps in their reasoning visible? Fourth, critical evaluation: did they check and challenge the AI output? Fifth, human contribution: what did they add that AI could not provide? Sixth, metacognitive accountability: can they reflect honestly on how AI shaped their work?
This approach does not ban AI. It does something more useful: it separates responsible AI-supported learning from lazy automation. It allows providers to say: “Yes, use AI — but use it to strengthen your thinking, not replace it.”
This is especially urgent for South African education and training. We are already under pressure on quality, employability, workplace competence, assessment credibility and regulatory trust. If AI becomes a shortcut around learning, the system weakens. But if AI becomes a tool for deeper questioning, better feedback, stronger reflection and improved learner support, it can be a powerful ally.
Cognitive Authenticity therefore offers a more mature way forward. It moves us beyond fear and towards responsible design. It encourages providers to redesign assessments, include oral defence, request declarations of AI use, assess drafts and evidence, and ask learners to explain their thinking. In workplace learning, it can help confirm whether competence is genuine or merely well presented.
The future of education will not be AI-free. Nor should it be. But it must remain human-led. The goal is not to protect outdated assessment habits. The goal is to protect thinking itself.
In the age of AI, integrity is no longer only about whether the work is original. It is about whether the mind behind the work is still awake.
Cognitive Authenticity may become one of the most important quality concepts for education, training and assessment in the AI era.