People trust AI more when the problems get harder, which is exactly when it’s most likely to be wrong.
This 'verification paradox' reveals a 46-point gap between perceived and actual AI accuracy. Because humans are susceptible to 'automation bias,' they mistake the confident, fluent tone of a large language model for correctness precisely when the complexity of the task has caused the model's actual performance to plummet.
A Pedagogical Framework and Its First Classroom Implementation in Response to Automation Bias, Cognitive Debt, and the Verification Paradox
EdArXiv · vhwbn_v2
Generative AI (GenAI) has become cognitive infrastructure in higher education, yet creates a verification paradox: student reliance peaks where task complexity is highest, objective accuracy lowest, and perceived correctness remains inflated (46-point calibration gap). This paper presents the ACTIVE Framework, which is a six verification principles operationalized as a five-step workflow (Assess, Constrain, Inspect, Verify, Explain), and its first classroom implementation at Deggendorf Institute