Schools Are Repeating the Exact Mistakes That Wasted 15 Years of Web Literacy Education... This Time With AI
A cottage industry of consultants is currently traveling the UK and US selling AI literacy frameworks and teaching practices to desperate schools. These frameworks claim to prepare students for an AI-powered future. They look professional, evidence-based, credible.
They have, according to MIT Professor Justin Reich, "about as much evidential support as the CRAAP test did when it was invented."
If you're unfamiliar with the CRAAP test, the name alone should tell you how that turned out.
From 2003 onwards, millions of students were taught to evaluate websites using CRAAP—Currency, Reliability, Authority, Accuracy, Purpose. The method told them to closely read sites for citations and proper formatting, avoid Wikipedia, and trust .org and .edu domains over .com sites. Library science experts endorsed it. It seemed reasonable, evidence-informed, professional.
When rigorously tested in 2019, students using CRAAP performed miserably at sorting truth from fiction online. Experts used a completely different approach—quickly leaving suspicious pages to check how other sources characterize them, a method called "lateral reading." Schools had confidently taught the wrong thing for 15 years. The name, in retrospect, was prophetic.
Now, the digital media professor says we're repeating the pattern with AI but with a devastating twist.
Reich considers AI an arrival technology: it crashes the party uninvited, then rearranges the furniture. Unlike computer networks that required procurement decisions and were policed by a staff member, AI tools are already on every student's phone. Schools face a brutal paradox: rush to adopt unproven methods risking another painful mistake for young people, or risk dithering leaving students to navigate AI alone.
Teachers interviewed by Reich's MIT team share one desperate refrain: "don't make us go it alone."
Reich believes the evidence won't arrive until around 2035. By then, another generation will have been the experiment. Reich's advice? Humility, local experimentation, rigorous assessment of what actually works for pupils, not pinning hopes on overconfident frameworks sold by consultants who profit from urgency.
Schools don't need a race to adopt AI first. They need a race to get it right for young learners, before today's AI literacy frameworks earn an equally unfortunate acronym in the years to come.
Questions:
When AI crashes the party uninvited on students' phones, how do schools protect young learners without repeating past failures?
Who bears responsibility when today's AI literacy frameworks fail the next generation of students?
