Classroom experiment: students fail to spot misinformation from a confident AI chatbot

Working_Original9624 · reddit · 2026-11-02

A South Korean undergraduate indie developer took part in an education experiment featured in EBS's documentary The Extinction of Thinking, testing whether middle school students can recognize misinformation delivered by a confident AI chatbot in real history and ethics classes.

Setup

Teachers prepared false claims and scenarios in advance; two tools were built:
An AI misinformation chatbot that intentionally delivers misleading answers based on teacher-provided scenarios, with randomized difficulty levels;
A teacher dashboard to monitor students' conversations and responses during the activity.

The goal wasn't a better chatbot, but letting students experience how easily AI-generated information can shape their understanding.

Findings

Even while attending lessons, completing assignments, and discussing the material, participating students failed to identify the misinformation. The author cautions this was a classroom experiment, not a large-scale study, but raises a key question: are students learning to understand information or merely to obtain answers — and how do we teach critical thinking when AI can produce convincing explanations for untrue claims?

Open source

The classroom tools are open-sourced via NomaDamas on GitHub (EBS-Gurapingala-teacher). The experiment will appear in the EBS documentary premiering November 2, 2026.

Original post →

Classroom experiment: students fail to spot misinformation from a confident AI chatbot

More from AGI Musings