It’s not coming around the corner… it’s already here. Machine learning is going to have a profound and lasting effect on the investment landscape, as we know it now. Assets and procedures that we think of as permanent features are slowly disappearing. According to a late-2017 J.P. Morgan report, Big Data and AI Strategies: Machine Learning and Alternative Data Approach to Investing, the more expedient and accurate nature of the feedback that can be gained from a tech-enabled data source will eventually render traditional data sources – such as quarterly reports and other “low frequency macroeconomic data” – obsolete.

While on the one hand, saying goodbye to the quarterly report might be painful to imagine, the trend is undoubtedly positive, which is good because it’s also unavoidable. Read on to understand how machine learning in finance is driving improvements to both internal processes and banks’ and other financial institutions’ abilities to meet the needs of their customers.

Machine learning vs. AI

While the terms ‘machine learning’ and ‘AI (artificial intelligence)’ are often used interchangeably, they do not mean exactly the same thing. The broader of the two is AI, referring to any technology that enables a system or device to make intelligence-based decisions or actions. Machine learning, on the other hand, refers to the underlying mathematical models and algorithms which allow for the automated utilization of data to make decisions.

With hundreds of millions of customers—ranging from individual account holders to vast enterprises—the financial services industry generates massive amounts of data. Machine learning offers banks and other financial institutions the potential to harness the actionable intelligence that exists within this information, delivering tangible benefits that range from improving risk control to seamlessly fulfilling customer needs.

Better fraud detection and mitigation of other risks

While the potential of machine learning and AI—specific to the financial industry are still being analyzed, a report from McKinsey estimates that its full application could reduce risk-related costs for global systemically important banks by up to $1-billion annually; for domestic systemically important banks that number is a not-too-shabby $400-million in annual savings.

Machine learning can help financial institutions both avert losses and reduce the cost of doing so in a number of different ways, such as:

  • Thwarting fraud

From credit card fraud and the opening of fake accounts to identifying sham financial services professionals, machine learning can help financial institutions identify potential fraud more efficiently and quickly. An analysis of how the Australian Securities and Investments Commission uses machine learning to thwart fraud found that automated data analysis of the marketing materials of rogue financial advisers proved five times better than random checks for finding “language that merits referral to enforcement.”

  • Identifying high-risk customers

A report by the rating agency Moody’s found that machine learning models are more accurately able to predict whether a borrower will be able to meet loan payments than a human using and interpreting the results of a proven statistical model. This means that by incorporating machine learning, lenders will be able to more effectively protect themselves from undue losses, charge more risk-appropriate interest rates, and improve the speed and accuracy of their entire credit application process.

  • Improving regulatory and compliance reports

High levels of oversight, related to the Dodd Frank Wall Street Reform and Consumer Protection Acts and other regulations within the financial services sector, place a steep burden on banks, forcing them to account for hundreds of millions of contracts and legal documents. But to cite just one example of the potential for machine learning on the compliance front, JP Morgan Chase reported that the introduction of a machine learning platform helped them to shrink the manual review of 12,000 annual commercial credit agreements from 360,000 hours down to mere seconds.

  • Improved Customer Satisfaction

The adoption of machine learning has yet another potential advantage: helping banks deliver more efficient and intuitive customer service that better connects existing and prospective clients to products that will suit their needs. This in turn has the ability to drive customer loyalty and satisfaction, thereby ultimately driving revenue. The McKinsey report cites a top consumer bank in Asia that utilized customer data to define 15,000 micro segments within its customer base, using the resulting data to develop a next-product-to-buy model that increased the likelihood of signing up for a new product three times over.

Adoption is lagging

Despite its potential, research shows that adoption of machine learning in financial services is lagging. According to one report, less than one third of financial services firms report using cognitive technologies such as predictive analytics, recommendation engines, and voice recognition and response.

While general skepticism is one thing standing in many firms’ way, data silos—which limit the ability of new technologies to extract meaningful insights from the information being collected—are another commonly-cited challenge. Furthermore, the McKinsey report’s identification of factors such as “failing to have a clear view at the outset for the expected use or returns of machine learning” and “not asking the right questions, so that algorithms don’t deliver actionable insights”, reflect the importance of kicking off any initiative around machine learning in financial services with a holistic plan, first for transforming data into an ingestible format and then for extracting meaning from the vast stores that exist.

Wrap up

The slow adoption of machine learning within the financial services industry means banks are still just beginning to realize the improved efficiencies and cost savings machine learning has to offer. But in order to see these transformational benefits—from risk mitigation to better customer satisfaction—firms must first develop holistic plans for transforming and leveraging the vast amounts of data they possess.