Transforming India from Consumers to Producers of
AI models

As a nation, we find ourselves primarily as consumers of third-party AI models rather than creators. For example, why haven't we developed an AI model similar to Med-PaLM, customised to address the unique needs of Indian patients? Or, why do our underwriting and fraud-detection models remain rudimentary compared to global standards?

Let's pause and think. For a thriving AI ecosystem, four key elements are essential: a) users of AI models, b) tech talent to develop models, c) computing power to run AI models, and d) real-world datasets for training AI models.

In India, the accessibility of these training datasets poses a significant challenge. However, are these datasets available? Yes! For instance, consider the vast troves of finance data held by banks or health records maintained by hospitals—numerous datasets exist but remain locked away. Even when these institutions consider sharing their data for monetization, they encounter two major hurdles: a) maintaining customer privacy in compliance with data privacy laws (DPDP, GDPR, etc.), and b) retaining end-to-end control of their data at all times.

This is the challenge we aim to address. At PeriData, we have developed a clean-room software atop the techno-legal framework, DEPA Training, initiated by iSPIRT. This software facilitates seamless and secure access to and sharing of datasets among organizations while mitigating the risks of data privacy breaches and ensuring compliance with privacy regulations.

Our objective is to establish a platform for secure data sharing among organizations, fostering India's transition from being consumers to producers of AI models.