Human-Centered AI and Explainability

Overview
In an era where AI systems increasingly influence decisions in healthcare, finance, and governance, technical accuracy alone is not enough. We focus on Human-Centered AI (HCAI), ensuring that automated systems are understandable, trustworthy, and aligned with human values. Our work in Explainable AI (XAI) goes beyond "black box" models to create interfaces that help users, whether doctors, policymakers, or citizens understand why an AI made a decision, especially in high-stakes contexts.

Key Focus Areas

• Interpretable Machine Learning: Developing models that provide transparent reasoning for their predictions, particularly for local languages (e.g., Bengali NLP) and critical services.
  
   
• Trust & Calibration: Studying how users perceive AI reliability and designing interfaces that prevent over-reliance on flawed systems.
  
   
• Human-in-the-Loop Systems: Designing workflows where human expertise augments AI capabilities, ensuring accountability in automated decision-making.
  
   
• Responsible AI Deployment: Evaluating the ethical implications of AI tools in the Global South, checking for bias, fairness, and unintended social consequences.
  
   

Our Impact

By making AI transparent and accountable, we aim to build systems that people can confidently adopt and control, reducing the "trust gap" between advanced algorithms and the communities they serve