Tokenization of Digital Assets for Ethical and Equitable Model Training in an AI-Based Proctoring Application

The COVID-19 pandemic catalyzed a significant shift in education towards online learning and remote examination processes. AI-based proctoring tools emerged to maintain exam integrity but raised ethical, bias, and privacy concerns. This paper proposes a blockchain-driven architecture to address these issues. Our approach incentivizes genuine data collection through blockchain-powered crowdsourcing, rewarding contributors with crypto tokens. It ensures decentralization by using a voting system to validate labels given by annotators for user images, with winners also receiving tokens. Emphasizing user privacy, our model enables users to retract image access, triggering server image removal and model retraining. To enhance accessibility, our trained models are available as a service through layered APIs accessed via crypto tokens. This proposed architecture aims to reconcile the challenges of online proctoring while prioritizing ethics, fairness, data reliability, and inclusivity.