This study evaluates the effectiveness of ADAI, an interactive android-based teaching assistant, in fostering trust and reducing mistrust among engineering students during micro-classes. ADAI integrates computer vision, natural language processing, and speech synthesis to deliver adaptive explanations, detect student participation, and administer assessments via QR codes. A mixed-methods approach assessed trust and mistrust after student interaction with the system, using validated instruments focused on reliability, safety, and perceived risk. Statistical analyses included Shapiro-Wilk tests, Pearson correlations, and paired-samples t-tests. Results showed significant correlations between trust and mistrust dimensions, indicating interdependence between these perceptions. Paired-samples t-tests revealed significantly higher trust than mistrust scores, reflecting a generally positive reception of ADAI. Although exposure to robotics slightly increased perceived mistrust compared to non-exposure, trust levels remained stable, and no gender-based differences were observed. Overall, the findings suggest that ADAI supports interactive learning while maintaining positive perceptions of reliability and safety, highlighting its potential for enhancing engagement and trust in engineering education.
