It was one of those rare cold mornings in Bangalore in year 2027, the construction dust was floating in the air and visibility was close to 0 meters in Whitefield. It wasn’t even 9AM and the roads were already jammed on the way to ITPL. Software Engineers were rushing to their work on newly opened Metro while the bold ones decided to drive to work that day.
ITPL, which houses many of the multinationals and successful Indian banks, had its parking lot overflowing already. Country’s 4th largest bank had a technology support and analytics team in ITPL. Over past decade, their technology support team grew while the analytics team shrunk year over year.
Analytics team had 3 people, a manager and 2 analysts who withstood waves of layoffs in the past few years owing to advancement in AI and Machine Learning. The 3 people analytics team was only focusing on liaising with technology support team in keeping analytics server and applications up and running. AI took decisions on which, when, how to run ML models. AI took output from these ML models and drove business decisions.
One such AI module was responsible for refreshing ML models for credit scoring, scoring incoming applications and making approval decisions on loan applications and deciding on credit limit and interest rate. The ML models were so sophisticated that they could predict with 98% accuracy the probability of default and the net present value of an applicant. The bank was heavily reliant on analytics to ensure that they onboard the right customers and grow profitably. It was just a few years ago that they had a set of super smart computer programmers and statisticians who collaborated to build this AI and ML framework. But they were let go because of automation of their jobs achieved through the intelligent system that they built in the first place.
AI and ML was not just common in companies but it had penetrated into common man’s life too through mobile apps. There were apps that would make financial decisions for an individual using AI and ML algorithms. The algorithms were so sophisticated that it knew how to manage credit to get the maximum credit score.
Over 100,000 people subscribe to our newsletter.
See stories of Analytics and AI in your inbox.
The % of applicants who used this AI & ML powered mobile app had been increasing over years. This made it very difficult for the AI and ML credit scoring algorithms to differentiate among applicants when most of them had near perfect credit history. On that particular day, the bank received 1500 applications and all of them where from people who used AI & ML powered mobile apps. The credit scoring algorithm scored all applicants and decided that everyone qualified for a loan with the highest credit limit and lowest interest rate. A quality monitoring AI algorithm triggered an alarm and alerted leadership team about the potential issue in credit scoring decision. Leadership team reviewed logs by working with the 3-member analytics team and got thoroughly confused with what was happening.
Customers used AI and ML to perfect their credit and the AI and ML used for making credit decisions at the bank scored all applicants as perfect applicants. The perplexed leadership team and the not-so skilled analytics team didn’t know how to fix this issue. The leadership team decided to turn down all the approved applicants through a personalized apology email. In the evening de-brief meeting, they discussed how to fix this bizarre situation. One of the managers screamed “Our machines are being defeated by customer’s machines. The battle of AIs have begun”. Everyone in the room went into deep thoughts on what they have done to their analytics systems.
It was truly a day when Analytics stood still.