3 Reasons Why Banks Can't Afford to Ignore AI

The promise of Big Data and our ability to use it was a lofty one. Unfortunately, many Big Data efforts have turned out to be extremely costly, both in terms of the investment and the human capital required to draw insight out of it.

While companies use many approaches to extract value from data they capture, organize and store, it is only recently that we have seen a new class of technologies emerge providing real impact — those powered by Artificial Intelligence, aka AI.

No matter what you call it — “cognitive computing,” “machine learning,” “cognitive analysis,” “smart machines” or even “intelligent automation” — AI is enjoying a renaissance now, not simply because of the promise it holds for the future but because of the impact it is having on businesses today.

In banking in particular, where data drives decision-making, reporting and customer communication, AI applications are already transforming the way business is getting done. The three areas where this impact is most recognizable are reporting, advising and alerting.

Communication and reporting has always been a focus for banks. They employ significant resources to report on the performance of portfolios, client investment strategies, market conditions, fraud and compliance. Generating such reports is a time-consuming effort even for the most skilled and experienced analysts — and that is time that could be better spent elsewhere.

When it comes to reporting, we are seeing tremendous impact from AI in the form of systems that can instantaneously analyze data, reason about what that analysis means and then generate natural language reports that inform many audiences. Built on a combination of data analytics and reasoning systems, these new AI engines are able to bridge gaps between data that machines need for analysis and the language-based communication that humans need to understand it and take action.

Money laundering, fraud and compliance reporting have always taxed banking resources, and to mitigate costs banks have come to rely more on technology-based fraud detection systems. The issues with such systems are twofold – 1) they have to be responsive to new approaches fraudsters are taking to hide irregularities, and 2) they need to be able to recognize patterns of irregularity using both structured and unstructured data.

For the former, AI systems based on machine learning techniques can predict irregular activity. In much the same way that machine learning has been used for years to improve algorithmic trading systems, it is now being used to protect our financial system.

For the latter, natural language processing (NLP) systems are uncovering subtle cues in transactions that might indicate behaviors not seen in the numbers but seen in text. So, here the AI is aimed at understanding what the text means and transforming it into structured data. Although in the early stages, it is already improving organizations’ ability to recognize and respond to problems at scale.

We have also seen AI systems offering client-facing advice. On the small side, wallet.AI is beginning to provide advice to end users based on an analysis of their personal finance data combined with an understanding of goals. Advice is driven by the system having a “conceptual awareness” of client goals and needs. On a larger scale, USAA is deploying an investment advisory system for veterans using IBM’s Watson. Based on a combination of large scale data repositories and Watson’s “cognitive computing” model, USAA will be using the systems to give veterans financial advice related to their transition from military to civilian life.

Both of these models are based on the concept that some advice is a product of data and best practice rather than deep expertise. For wallet.AI, the system analyzes personal financial data to create a user profile. Combined with other information, that profile is used to drive a best practice advisory engine. USAA has a data set consisting of best practice documents, and Watson is the intelligent access engine that maps information about a client to the advice that best serves them.

The world of data and our ability to deal with it at scale has always been highly mechanical, which has limited our ability to extract all of the value we need. We are now entering an age where we can combine the human skills of interpretation, communication and language understanding with the machine’s ability to scale, resulting in powerful systems that allow us to focus on more strategic work and provide meaningful information to any audience.

In addition to being Narrative Science’s Chief Scientist, Kris Hammond is a professor of Computer Science and Journalism at Northwestern University. Prior to joining the faculty at Northwestern, Kris founded the University of Chicago’s Artificial Intelligence Laboratory. His research has been primarily focused on artificial intelligence, machine-generated content and context-driven information systems. Kris currently sits on a United Nations policy committee run by the United Nations Institute for Disarmament Research (UNIDIR). He received his PhD from Yale.