Developer finds four conflicting grade engines in AI-built education app
A software developer was brought in to make a production-ready an education platform that had been built using Lovable, an AI-powered 'vibe coding' tool, over several months. An automated audit of the exported codebase returned 319 issues, including 29 critical ones, among them a flaw allowing students to grant themselves administrator privileges. The most serious problem discovered was four separate, independently functioning engines all calculating the same daily student score and writing results to the same database field — with conflicting weightings that produced different grades for the same input. Because no commit history, pull request notes, or documentation existed to indicate which engine was authoritative, the developer could not resolve the conflict without risking changing students' actual grades. The issue was escalated back to the client as a product decision, highlighting a key risk of AI-generated codebases: readable but unexplained logic with no traceable reasoning.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in