The AI-Literate Learner Can Refuse the Tool

A learner opens an artificial-intelligence tool, recognises that the task involves confidential information and closes it. Another checks a confident answer against an authoritative source and rejects it. A third asks for a human explanation because the automated feedback cannot show how the decision was reached.

These are not failures to use technology. They are signs of literacy.

Much AI education still begins with access and ends with prompting. Learners are shown how to produce text, images, summaries or code, then described as future-ready because they can make a system respond. But fluency with an interface is not the same as informed judgement about a sociotechnical system whose output can be persuasive, wrong, biased, invasive or inappropriate to the task.

The AI-literate learner can use the tool. The more important learner can also verify it, limit it, disclose it, escalate beyond it and refuse it.

On 21 August, Associated Press reported that public-school systems in the United States are increasingly moving from blanket chatbot bans towards AI-literacy policies and training. In Charleston County, South Carolina, a policy developed with teacher, learner and parent input has entered a second phase involving training for teachers and middle- and high-school learners. The training includes visible system failures, source verification and data privacy.

The report says thirty-seven states have issued guidance. Utah has trained more than seven thousand teachers…