Harender Bisht

Explainable AI for High-Stakes Decision Analytics: Building Trust, Fairness, and Accountability in Insurance Fraud Detection

Abstract:

Artificial intelligence is increasingly adopted in high-stakes decision-making environments, such as insurance fraud detection, financial risk analysis, and compliance monitoring. While AI models can improve detection speed and analytical accuracy, their outputs must also be explainable, fair, auditable, and actionable for business users, investigators, and regulators.
This talk focuses on how Explainable AI can be designed for real-world insurance fraud detection workflows. It discusses how AI-supported decisions can be connected to evidence, reviewed by human analysts, governed through fairness and accountability controls, and documented for auditability. The session also highlights practical architecture patterns such as human-in-the-loop review, confidence-based escalation, model governance, and responsible AI controls.

The goal is to show how organizations can move beyond black-box fraud scores and build trustworthy AI decision systems that support transparency, compliance, and responsible adoption.

Profile:

Harender Bisht is a Solution Architect and AI researcher with professional experience across enterprise systems, finance technology, insurance analytics, data engineering, and responsible AI. His research focuses on Explainable AI, ethical AI governance, fairness, accountability, transparency, and the application of AI in insurance fraud detection and high-stakes decision analytics.
He has worked on enterprise data platforms, reporting systems, finance applications, and AI-enabled analytics solutions. His work bridges academic research and enterprise implementation, focusing on building AI systems that are technically effective, auditable, trustworthy, and usable by business stakeholders.