Machine Learning Helps Credit Unions Become More Personal, Secure
The irony is not lost on how nonhuman interaction using machine-learning capabilities in financial technology could supply a more personalized member experience, as well as added security and operations performance.
The usage of AI, and its subset machine learning, is a rising trend among financial institutions as they seek to improve customer satisfaction, reduce inefficiencies and fight fraud.
Jerry Melnick, president/CEO of the San Mateo, Calif.-based machine-learning analytics firm SIOS, defined machine learning as a type of artificial intelligence and method of data analysis that uses algorithms to draw conclusions, make predictions or learn without additional programming.
AI runs the gamut from voice recognition to neuronetworks to chatbots, Terence Davis, vice president and chief architect at the Santa Clara, Calif.-based Incedo said.
“One of the subgroups is this machine learning. We focus on innovation and the data lifecycle enabled by a revolution in hardware capabilities and technology around data,” Davis explained.
Incedo typically helps clients determine what they want to extract from their data. “Organizations need to know how they want to use the data before actually molding the data for their use,” Davis stated. “It's not quite at the level of Star Trek where you can just talk to the machine and say, ‘Computer, collate this data.’ You need to put some thought in ahead of time.”
One of the drivers encouraging the integration of machine learning in financial services organizations is cost reduction. If Incedo can implement some sort of animated chatbot, for instance, it can enable human service representatives to focus on higher value calls, Davis said.
Davis noted one of the important benefits of using AI and machine learning is data quality. “If you want to gain some of these deep insights, if you want to answer questions for your customers or executive team based on deep learning and machine learning, you need to have high quality data,” he said.
Credit unions are no different from many other organizations. “They have a lot of data in different forms, a lot of different standards being used, so they are going to have to face this data quality issue just the same as very big banks,” Davis said.
AI capabilities through integrated natural-language processing engines and chatbots can transform the mobile app user experience.
“Chatbots are one of the more popular of the artificial intelligence technologies because they enable a conversational experience for customers,” Antonio Sanchez, product marketing manager at the Austin, Texas-based mobility solutions company Kony, said.
Kony believes chatbots can connect members with artificially intelligent service representatives to check account balances, report lost cards and complete other tasks such as advanced portfolio analysis.
Sanchez explained chatbots can do the following:
Interact via conversation or text;
Take users to the area within the app that allows them to complete functions quickly;
Process what users want to do; and
Minimize the number of taps or swipes a member would normally make.
Kony does not set up its own bots; instead, its architecture allows most natural language processors or chatbot services to plug into its banking app. There are a number of different types of chatbots in the financial sphere. With Kony, credit unions provide the application program interface and can then map into Kony's middleware layer.
Sanchez explained, “Part of artificial intelligence is predictive analytics, which you can say is the same as machine learning.” This is where organizations take available data to create a behavioral profile.
“When we talk to customers, they want to maintain control of the digital experience so they prefer that the chatbot be integrated as part of the app,” Sanchez said. “This goes to providing that personalized experience.”
Sanchez maintained that right now, machine learning and chatbots are some of the “very cool, bleeding edge, innovative” types of functionalities. “But like anything else, it's not going to be long before that kind of stuff is just going to be table stakes just to be able to compete.”
Another way credit unions can use machine learning is for fraud monitoring.
The Hoboken, N.J.-based NICE Actimize's new product, ActimizeWatch, is a cloud-based analytics optimization solution, which leverages machine learning and the cloud to provide proactive fraud analytics optimization and consortium data sharing.
“What we do is monitor the performance of financial institutions’ fraud analytics. We are able to change or optimize analytics very quickly and deliver them back with the ultimate goal of making fraud detection solutions and financial institutions as agile as the fraudsters,” Rivka Gewirtz Little, director of fraud product marketing for NICE Actimize, said.
She added, “Essentially all of our customers, whether credit unions or Tier 1's [banks], face ever-changing fraud. Keeping analytics ahead of those attacks is the Holy Grail for fraud detection.”
Some key points addressed by ActimizeWatch include:
The intersection of machine learning and the cloud enables financial institutions to share anonymized data and gain a cross-market view of fraud patterns.
Obtaining a cross-market fraud view allows institutions to stop threats earlier.
Continuous monitoring of transactional data in the cloud enables proactive fraud analytics optimization.
“All of our customers, whether they have an on premise or cloud solution, continually send us transactional data and we continually monitor threat data, identifying fraud patterns and using machine learning to optimize analytics very quickly,” Little said. “It's really interesting to look at it from the credit union perspective. There was a misconception that it required a team of data scientists to manage this machine-learning environment. ActimizeWatch promises the exact opposite for smaller organizations.”
Machine learning can also help financial institutions keep business-critical applications up and running. “When end users experience slow application performance in front-office operations, IT needs a way to identify and correct the root cause of the problem as quickly as possible,” Melnick said.
SIOS iQ uses machine and deep learning technology to improve IT operations analysis, optimization and performance resolution. It acquires a broad set of data in real-time across the infrastructure − CPU, storage, network, applications − and then applies machine-learning analytics to automatically identify abnormal behavior, its root cause and a recommended solution.
“This enables IT to move from being reactive in their approach to detecting performance issues to being a proactive, strategic and innovative part of the business,” Melnick explained.
This is especially vital with IT environments becoming increasingly complex as organizations move from physical servers to virtual environments comprising virtual machines, applications, storage and networks that are highly interrelated and constantly changing.
“Rather than monitoring individual components in the traditional way, artificial intelligence, more specifically, machine learning and deep learning tools, automate problem-solving by analyzing the behavior of interrelated components,” Melnick added.
Overall, seven out of 10 consumers around the world would welcome robo-advisory services for their banking, insurance and retirement planning, according to a report by Accenture. Yet, many consumers still prefer human interaction for more intricate requirements, leaving many financial institutions weighing how to incorporate computer-generated and human services.
“When we talk to prospects and clients, we recommend they need to start looking at artificial intelligence, predictive analytics and machine learning,” Sanchez said. He added many credit unions and banks look for ways to differentiate themselves and provide that unique personalized experience. “This is a great way to be able to do that. Financial institutions need to get on board sooner rather than later or risk being left behind.”