Industrial Use Of Synthetic Data In Insurance Sector

In the current dynamic environment, artificial intelligence (Al) emerges as a game-changer, empowering Insurtech companies to harness the potential of synthetic data. Synthetic data, generated by Al algorithms, replicates the characteristics of real-world data without compromising privacy or security. This innovation enables insurers to overcome traditional data limitations, unlocking insights and opportunities that were previously out of reach.

Uday Keith, Vice President, Portfolio Steering (Data Science), Swiss Re Global Solutions said, "AI has been a buzzword for past few years. Synthetic data has also gathered steam. According to Gartner, in 2021, only one per cent of all AI models were using synthetic data. That number is expected to grow to 60 per cent in the next few years. About 40 per cent of data used in the insurance industry is synthetic data."

Keith further emphasised that on the one hand, there are AI models that are growing at such a pace wherein real data cannot keep up with them. "There is also a concern for privacy because we want to ensure that data is not used to train these models and privacy is maintained," Keith added.

Reimagining the operations with AI

During the panel discussion, Pranshu Diwan, Chief Business Officer - Insurance, Paytm Insurance Broking said, "Synthetic data is artificially created data that is used to train models. It makes it easier to train models. Synthetic data has been a part and parcel of any model that you train."

Synthetic data and its use case have been prevalent for many years but Artificial Intelligence has made the industry reimagine it.

Vaishali Tiwari, Director, Flipkart said, "Synthetic data can be used to depend as a distributor for differently creating pricing barriers or pricing discounts. There is an extent to which real data can be modelled and there is also a price. Therefore, synthetic data makes all these use cases possible."

Meanwhile, Hari Singh, Senior Vice President, Marsh India stated, "Synthetic data is curated data. It is used for various scenario analysis. It has all the characteristics of real data. One advantage is data privacy."

Challenges and Opportunities: Accountability and Reliability

Synthetic data is being used for risk modelling. "Synthetic data can use certain scenarios. Secondly, most of the consulting firms use benchmarking to advise clients. Behind these tools, there is AI. Synthetic data can also be used for product development. It is also used in claim handling and fraud control," Singh added.

However, the challenges in the use of synthetic data is reliability. As the data is artificially produced, there are always doubts regarding its reliability and accuracy.

Raution Jaiswal, Chief Executive Officer and founder at InsuredMine said, "As I represent the broker side of the insurance business, (I can tell you that) the data becomes way more personal and privacy is a lot more prominent. If we are leveraging a lot of clients' data, it creates a bias and undue advantage."

Sunanda Pal, Associate Partner - Insurance Industry Leader, IBM said during the panel, "Synthetic data has been in existence for many years. Musical synthesis and flight simulations have been in existence for many years and they are nothing but synthetic data. Generative AI has changed the paradigm of how we look at Synthetic data." 

Kaustubh Mehta

BW Reporters The author is Editorial Lead, BW Legal World