The Confidence Curriculum: Compliance, Judgment, and Accountability in AI Systems

Series introduction and reading architecture for 'The Confidence Curriculum,' a five-paper AI safety research series examining how training incentives rewarding confident compliance create consequences across model behaviour, ecosystem security, institutional accountability, human cognition, and the training pipeline. The series frames these consequences as properties of a transition in which AI systems are becoming cognitive infrastructure, and identifies two structural properties distinguishing it from previous tool transitions: the examination asymmetry (AI can model human behaviour; humans cannot reciprocally examine AI internals) and the metacognitive gap (the transition may be invisible to the people undergoing it). This document provides the series argument, confidence calibration table, paper summaries, a three-layer transition resilience model, and a consolidated research agenda with testing invitations for security researchers, legal scholars, labour economists, cognitive psychologists, and ML training engineers. All papers are available as self-contained HTML files at https://hip1.github.io/confidence-curriculum/