Abstract

The integration of artificial intelligence (AI) into workplace environments presents both meaningful opportunities and complex challenges, particularly within highly regulated industries such as banking. While AI has the potential to improve efficiency, productivity, and decision-making significantly, its adoption also raises important concerns related to employee well-being, data privacy, and system security.

Through direct engagement with TD Bank, this project identified a critical gap: employees often lack the clarity, confidence, and trust needed to effectively and responsibly use AI tools in their daily workflows. This uncertainty does not only limit the potential benefits of AI but also introduces operational inefficiencies and increased institutional risk.

In response to this challenge, our team developed a Human-Centered AI Assurance Framework, supported by an AI-powered decision-support minimum viable product (MVP) website. This solution is designed to guide employees in real-time by helping them identify potential AI-related violations, understand associated risks, and make informed revisions to their actions. By combining structured governance with intuitive and accessible user interaction, the framework transforms static policy documents into dynamic, actionable support tools that align with real workplace needs.

Preliminary testing and validation suggest that this approach improves efficiency, reduces confusion, and enhances employee confidence when interacting with AI systems. More broadly, the project contributes to the ongoing conversation around responsible AI integration by emphasizing the importance of balancing technical safeguards with human-centered design. In doing so, it offers a scalable and practical model for organizations seeking to implement AI in ways that are not only secure and compliant, but also supportive of employee experience and well-being.