Chennai: A recent study by IIT Madras and Vidhi Centre for Legal Policy calls for a participatory approach to Artificial Intelligence (AI) governance, emphasizing inclusivity in development to enhance AI fairness and accountability.

Conducted by the Centre for Responsible AI at IIT Madras, the research suggests involving diverse stakeholders across the AI lifecycle—from design to deployment—to improve transparency, reduce bias, and foster public trust in AI systems.

The study highlights how AI's growing role in sectors like healthcare and law enforcement presents unique risks if left unchecked. In particular, Facial Recognition Technology (FRT) in law enforcement may perpetuate biases without proper oversight, raising privacy concerns.

Large Language Models (LLMs) in healthcare, meanwhile, could spread misinformation if trained on biased data, affecting medical decision-making. The researchers advocate involving affected communities, including legal experts and civil society, to develop fairer, more transparent AI applications.

Recommendations include adopting clear frameworks for stakeholder engagement, developing methods to incorporate feedback effectively, and maintaining human oversight, particularly in sensitive areas.

Professor B. Ravindran, Head of the Wadhwani School of Data Science at IIT Madras, stressed that responsible AI development requires accountability and ethics throughout.

The study, published on arXiv, also outlines tools like the "decision sieve" model to streamline stakeholder identification, essential for building equitable AI systems.

This participatory framework, the study argues, aligns with IndiaAI's goal of human-centric AI, ensuring AI systems are safe, reliable, and serve the broader public interest.