Trusted AI - when data cannot be trusted

Trusted AI is the trending research area. It is the need of the hour to showcase the bias and unfairness in the result from AI solution. But what if the data is riddled with inherent bias and unfairness even after demographic parity. What about the scenarios when human bias reflects in data labeling. What do we do in such scenarios and how do we create a Fair, reliable and trusted AI solution

 

Sandhya is an AI &MLOps practice leader at EY GDS. As organizations struggle to scale AI and realize business impact, her role is to conceptualize, build and deploy AI, ML @scale for EY Clients across sectors globally. She has scaled MLOps solutions and built target architecture working with client enterprise and data architects, CDOs and Businesses, which have enabled data-driven transformation for large Financial Services Clients. She has overall 17+ years of experience in the field of AI & ML. She leads go to market strategies working with EY globally on positioning of the AI and ML capabilities, evaluating vendors and partners to build an end-to-end solution architecture which have enabled multiple deals in this area of work with multiple sectors like global banks and asset management companies. She has expertise across cloud platforms and have developed end to end MLOps architecture on Azure, AWS and GCP for EY clients. These are positioned as logical and technical architecture for clients. This has enabled clients to productionize more than 30+ AI use cased on Cloud in areas of customer, risk, finance and operations.