All eyes on … AI in banking part 2/2
All eyes on … AI in banking part 2/2
Do You remember artificial intelligence from the movie “I Robot”? When character played by Will Smith asked the question “Can a robot write a symphony?” Instead of the original “Can You?”, todays’ Robots will be able to… write a symphony*. The development of artificial intelligence occurs very quickly, and banks as highly technologically advanced and managing huge amounts of data are natural candidates for using artificial intelligence to meet their clients needs.
So how can financial industry can use the potential of this technology?
The portfolio management process requires a lot of time and analysis of large amounts of data. Machine learning taking over this process is able to take into account individual factors, for example personal customer goals or individual risk tolerance. By analyzing and evaluating the available information, the algorithm is able to develop a portfolio approaching each client individually.
Credit risk assessment
Machine Learning allows analytics to get a much more accurate and multidimensional analysis of credit risk. Machine learning based on the patterns and trends extracted from a large amount of historical data effectively predicts the behavior of clients and uses all available and newly received information to predict the risk of customer insolvency. What are advantages of risk assessment supported by Machine Learning? Among others: speed of the process, more precise results (based on more data than is possible in the case of human analysis) or possibility to add client’s business environment context.
Additional advantages – as in all cases of Machine Learning implementation – the assessment can be based on any number of conditions/factors, both external and internal, and its effectiveness increases with increasing diversity of factors. Besides, the model ifself is verified and repaired by working, so if You let it operates longer the better results you can record.
Chatbots are not only machines that return planned answers, but also tools, based on subsequent conversations and information provided by us, gain experience, and in every next interaction provides services at a higher level. What are the advantages of such solutions, which are increasingly replacing channels such as email and telephone? Chatbot gives us a quick and personalized response and at the same time helps reduce cost. Therefore, it meets the needs of increasingly demanding bank customers who expect the bank to be available, in any place, any time, 24 hours a day, 7 days a week.
Also, in the case of advanced applications, integration with other channels allows access to all customer data. Thanks to this, the algorithm can track spending habits, help to set budgets, thus being a useful support for the user in money management.
Are these theoretical examples already working in real? COiN (Contract Intelligence) is a system that is mentioned as one of the most spectacular successes of Machine Learning implementation in the financial industry. The program is used by JP Morgan in analyzing legal documents and extract important data and clauses. What can illustrate the scale of the system’s usability? 12,000 contracts for manual inspection are equal to 360,000 working hours for JP Morgan. According to preliminary tests of the implementation of this COiN technology, with the help of Machine Learning, this work will be done in a few seconds, which will directly translate into huge savings for the bank. Wells Fargo and a chatbot project which interact with clients via Messenger, Bank of America and its virtual assistant Erica, who is a financial advisor to over 45 million bank clients are just a few proofs that the theoretical possibilities of using artificial intelligence are no longer just a theory.