Thu Nov 07 16:27:16 UTC 2024: ## IIT-Madras Study Advocates Participatory Approach to AI Governance

**Chennai, November 7, 2024:** A new study by the Indian Institute of Technology-Madras (IIT-M) proposes a participatory approach to governing artificial intelligence (AI), arguing that such a model would enhance transparency, foster public trust, and facilitate widespread AI adoption.

The research, conducted by the institute’s Centre for Responsible AI and the Vidhi Centre for Legal Policy, explores how involving relevant stakeholders in AI development can improve algorithmic outcomes and ensure fairness.

The study, published in the open-access archive arXiv, emphasizes the need for a collaborative effort, highlighting that those most impacted by AI systems often have limited input in their development.

“The study found that individuals most affected by the deployment of AI systems have little to no say in how they are developed,” said B. Ravindran, Head of the Wadhwani School of Data Science and AI (WSAI) at IIT-M. “A participatory approach would be beneficial in building and utilizing more responsible, safe, and human-centric AI systems.”

The study uses two case studies – law enforcement and healthcare – to demonstrate the benefits of a participatory approach. In law enforcement, the researchers argue that involving civil society groups, undertrials, and legal experts in the development of facial recognition technology (FRT) could lead to a fairer and more transparent system.

Similarly, in healthcare, engaging doctors, patients, legal teams, and developers in the creation of large language models could help mitigate the risks of inaccurate or biased information generated by these AI systems.

“Our report offers a sector-agnostic framework for identifying stakeholders, involving them throughout the AI lifecycle, and integrating their feedback,” said Shehnaz Ahmed, lead, law and technology at Vidhi Centre.

The study emphasizes that increasing transparency and accountability in AI systems fosters public trust, paving the way for broader acceptance of these technologies. By ensuring a more inclusive approach, the researchers believe that AI systems can become safer, more reliable, and less susceptible to bias and privacy violations.

Read More