A Course About AI Built With AI Powered By AI

Abstract

Abstract

For twenty-five years I wrote code by hand, four days out of five, to realize my educational visions. It was hard and often unpleasant—and that difficulty is, I suspect, why so many educators never build with the technology we tell our students can change the world. Since 2026 I no longer read, write, or debug code by hand, and I am building more than ever. This talk is what that made possible: Using and Understanding AI, a course for non-technical students with no prerequisites and no code, whose entire web platform was created by describing it to Claude in plain language across 199 sessions.

Summary

Summary

A ten-minute walk through one course, end to end, for an audience of Replit Faculty Fellows earlier in their own AI journeys. Three beats: the course (mental models of how AI systems work, plus the judgment to use them well), the collaboration (AI as a full co-instructor across content, infrastructure, and operations), and the assessment (conversational, multi-agent, and autograded—rigorous feedback at the speed of conversation). Along the way, raw Claude Code transcripts show what “building without code” actually looks like: a paragraph of plain English, typos and all, that became a working conversational oral exam the same afternoon. It closes on the hypothesis underneath all of it—that AI collaboration is a distinct skill, independent from but related to classical programming—and on the students, every one of whom chose to keep building with Replit when given a free choice.