A report by the Department of Economic Affairs, under the Finance Ministry, on fintech noted that fintech companies like Credit Mantri, CreditVidya and Samunnati were using AI and ML to create an alternate lending data score. It recommended that NABARD (National Bank for Agriculture and Rural Development) work along the same lines as these companies and create a credit registry for farmers for credit scoring, default analytics, predictive crop analytics, repayment monitoring, fraud control, and improving efficiency in credit services.
It’s worth noting that CreditVidya was found to be embedding tracking software in certain mobile apps, which gave it the access to a person’s GPS locations, and business SMSes from ecommerce sites and banks, according to a HuffPost India report. The user data was used to power CreditVidya’s algorithms that help lending companies determine the credit-worthiness of loan applicants. It ran this code for several months in 2017.
Here are some key recommendations made by the DEA surrounding AI and ML:
1. PSU banks can use AI and ML for risk management: The report recommended that the Department of Financial Services (DFS) and PSU (public sector undertaking) banks use AI, ML and cognitive analytics for risk management, compliance management and fraud control. It also said that the DFS and PSU banks should work on a roadmap to adopt these technologies in a time-bound manner.
2. AI can help in better compliance: AI has the potential to help firms detect compliance risks and make better-informed decisions about how to mitigate them, the report noted. It said that RegTech (regulatory technology) can reduce firm-level compliance risks and also reduce supervisory risks. The committee recommended that the financial sector regulators, such as RBI, SEBI, IRDAI, and PFRDA, must develop standards and use-cases for RegTech by financial sector service providers to make compliance with regulations easier, quicker and more automated for regulated entities.
3. Automation of compliance process: The FCA (Financial Conduct Authority) in the UK is experimenting with machine readable regulations, giving an impetus to the automation of compliance processes, the report said. It proposed creating a semantic web page – a web page similar to the current structure of the World Wide Web, but data and information on it can only be processed by a machine. The report said that AI can be used to construct such a web page which would in turn reduce the need for human intervention in compliance processes thus increasing the efficiency of the system.
4. AI can improve the cyber-security architecture: AI can be used by financial institutions including fintech firms to prevent cases of fraud and cyber-attacks, it notes. It highlighted that tools, systems and algorithms are being designed to detect cases of fraud or potentially fraudulent activities. “AI enables companies to track historical data of individuals through which any suspicious or outlier activity can be captured,” it added.
5. AI and regulatory compliance: Financial reporting has become more numerous and complex, the report noted. It added that this has placed upon companies an increased burden vis-a-vis compliance costs. “It has been noted in many quarters that AI could be harnessed to automate compliance processes which would enable saving of costs and time,” it said.
6. Auto-insurance claims can be made smarter as well with the use of AI and big data by use of image-based assessment smart insurance system, the committee said. It proposed that images of the accidental vehicle could be used and a claim amount could be calculated by tapping on a database of vehicle model(s) and part(s).