Responses from an Adaptive Trust-Calibrated Human–AI Collaboration for Enhanced System Usability

The dataset used in this study consists of participant-level interaction data retrieved from Human–AI collaboration outcomes under a controlled experimental design. It includes responses from 350 student participants (200 male and 150 female), each exposed to four system conditions: Static AI Assistance, Fully Autonomous AI, Rule-Based Adaptive Interface, and the proposed Trust-Calibrated Framework. The data resulted to 1,400 observation records. For each participant–system interaction, the dataset captures normalized metrics of task accuracy, trust calibration error, cognitive workload (NASA-TLX), and user satisfaction, alongside participant identifiers and gender information.