A bot that's confidently wrong — until your students correct it
HelpTheBot flips the usual AI-assistant script. Instead of students asking a bot for answers, they teach one. Each bot is built to hold a specific misconception about your course material and argue for it persuasively, in character — the student's job is to find the flaw and correct it with a precise, well-reasoned explanation before the bot will concede.
It's built for courses where the material is dense, the class is too large for small-group discussion of every misunderstanding, and students have started treating AI as a search engine that just agrees with them — quietly reinforcing well-argued misunderstandings instead of challenging them. HelpTheBot turns that around: a smart, deliberately flawed sparring partner that forces students to use the field's own vocabulary and defend their reasoning.
- The bot opens by asserting its misconception, in character.
- The student finds the flaw and corrects it — vague pushback doesn't count.
- Once genuinely convinced, the bot concedes and opens a follow-up question to extend the discussion.
Teachers author each bot's persona, misconception, and hidden grading criteria, then share one anonymous link with the class.