This repository contains code related to software to encourage humans to review software generated by AI.
The core premise is that rather than just asking humans to review AI-created or AI-modified software (and having humans may well just say they did after a quick skim due to seeing that the software works and lacking tume to review more thoroughly), that humans are incentivized to review more thoroughly by:
- giving them more 'points' when they make concrete recommendations such as adding a test or improving some aspect of the system,
- giving them even more points if they actually find various kinds of defects (especially ones related to security
This is currently in the early development phase.