Burdened with outdated legacy systems and the rise of advanced fintech banking options, banking executives continue to seek out new ways to boost margins and embrace the technological disruption prevalent in their industry. While the promise of artificial intelligence signals new opportunities for robust growth, it's important for banks to first ensure they have all their data organized and ready to use for the prerequisite machine learning necessary to ramp up an AI system.
AI takes over
While popular culture has been predicting the coming of advanced AI for decades, it's taken a bit longer for reality to catch up to the movie magic. But now AI is a major part of society and business, as consumers comfortably talk to their Amazon Echoes or easily discuss their daily schedules with Apple's Siri.
"Some analysts dubbed 2017 as the Year of AI in banking."
Beyond the personal assistant, another place AI has planted a firm foothold is the banking sector. This industry has seen such a prolific surge in AI implementation that some analysts have already dubbed 2017 as the Year of AI in banking, according to Bank Innovation. There's good reason for this appellation, as banking AI continues to surpass expectations and deliver complex transactions and analyses in a fraction of the time it takes humans.
For example, JP Morgan recently launched its Contract Intelligence program, Bloomberg Markets reported. COIN, as it's known, operates on a machine-learning system powered by the bank's private cloud network. In years past, the bank's lawyers and loan officers would spend up to 360,000 hours annually pouring over documents and data for routine tasks like analyzing commercial loan agreements. Following the bank's implementation of COIN, JP Morgan slashed this time down to mere seconds, going from 15,000 days – more than 41 years of collective work – to less than a minute. Further, COIN has also greatly reduced the mistakes that arise from human error.
With this kind of computational power at their disposal, it should be no surprise that banking executives and risk committees are clamoring to craft an AI strategy and start implementation as soon as possible. However, before launching any sort of banking AI – whether chatbots, automated services or even something that hasn't been invented yet – it's crucial that banks have their data ready.
Getting financial data in order first
AI will no doubt play a crucial role for financial institutions in the years to come, but any AI strategy must include a data organization phase from the onset. As an AI program analyzes new information, it compares the results to existing data to look for patterns, similarities and differences, thereby improving its ability to predict and classify new data.
Banks must first prepare their data for the artificial intelligence to utilize. Machine learning, the foundation of AI, requires substantial training and evolution on the part of the system. The AI needs to assess the bank's data, understand it and then make an appropriate decision based on what the bank would do. This process teaches the machine how to make the right choices. Without access to enough of the right kind of data, AI programs cannot reach their full learning potential, thereby negating their usefulness.
This means that to accomplish reliable AI machine learning, the system must have that data. Further, this data must be properly formatted, as well as consistent and accessible for the AI program to use.
One way banks have been responding to this shift toward data prioritization has been by restructuring with the goal of better data management. For instance, in 2014, Wells Fargo created the Chief Data Officer position to guide the bank's data strategy and manage risks, NerdWallet reported.
"AI is like a child, it needs so much information and requires so much data," Arif Ahmed, senior vice president of payments innovation for U.S. Bank, told Bank Innovation.
With more methodical data collection and better structured data caches, banking institutions can equip their AI programs with the information needed to optimize machine learning and truly propel their operations into the future of banking.