AI models must be transparent for better traceability, say experts

Experts from various sectors spoke on the ‘Ethical Considerations in AI: Navigating the Moral Landscape’ in a session moderated by John Xavier, Editor-Technology, The Hindu.

Highlighting how AI is used to maintain electronic medical records and assist in solutions to build diabetic retinopathy, Srinivasulu Thayam, Chief Technology Officer, Aravind Eye Care System, said there were challenges in addressing synthetic data. Besides validating the data used for decision support system, clinical stakeholders are explained about the outcomes. Moreover, patients too are informed about the use of data for research through AI system, he said.

While Balaji T.K., Assistant Vice-President, Engineering, FlexiLoans, noted that AI algorithms are used for credit scores for loans and its assistance is sought with data patterns for credit assessment, Balakrishna Kanniah, Chief Information Officer, VA Tech Wabag, noted that AI gave an insight into data analytics to develop a decision support system on water storage. It helped in preparing a gist of large tender documents and ensure quick decisions.

Kavitha V., Head, Department of Data Science and Business Systems, School of Computing, SRMIST, said that AI is a non-linear model and vigorous testing process would help arrive at a reliable model. Students are taught to come up with products and AI solutions where outcomes can be interpreted.

Panelists suggested that an alternative mechanism must be in place to audit AI system and ensure fundamental ethical policy for AI. The government has established guidelines for development and use of AI models.