Machine learning (ML) can crunch millions of data sets in the time it takes to withdraw cash from an ATM, making it a powerful tool for the financial services sector. It can detect fraud, automate trades, and complete banking and insurance tasks with minimal human intervention. What’s not to like?
Still, numerous challenges surround machine learning in finance. To help unpack the topic, DragonSpears interviewed Alex Muller, founder of end-to-end machine learning tool SAVVI AI. He shared his expert insights about the current state of this technology in finance and how opportunities to leverage it are beginning to open for smaller companies of all sizes.
What’s Going on With ML Right Now?
“Machine learning has been around for a while,” says Alex. “Historically, only mega-companies used it, but now everyone has some flavor of machine learning.”
People interact with ML daily, often without realizing it. Self-driving cars, Amazon and Netflix recommendations, airport video recognition — that’s machine learning in action, and it’s changing the world. Still, ML is in its infancy and, like with any technology, “it’s not fully honed yet.”
Impact of ML on Finance
Financial institutions could save $140 billion in productivity gains with ML, according to Accenture. In addition, McKinsey thinks AI technologies like machine learning could deliver up to $1 trillion of additional value. Whatever the number, ML could prove lucrative for any financial organization that wants to automate, streamline, and consolidate tasks.
From Alex’s perspective, the most successful implementation of ML in finance has been fraud detection. By using ML algorithms, financial services providers can identify theft and stolen credit cards and provide better services for customers. ML does this by analyzing a customer’s spending habits and triggering fraud warnings if it recognizes abnormal behaviors, such as a transaction at a gas station in California, followed by one in Paris.
ML has also had an enormous impact on money laundering and “know your customer (KYC)” — the process used by financial institutions to identify a customer and their risk potential. “Is Lisa really applying for a credit card?” asks Alex. “Machine learning can predict if it’s the right person by looking at the normal information like IP address, physical address, email, and phone number, but also at surprising information like time of day, voice patterns, keystrokes, GPS-based location, or even misspellings.”
Financial services providers use AI to predict patterns and understand actual and fraudulent behaviors. “If you change your credit card address and then order a new one, for instance, it’s more likely to trigger a fraud warning,” says Alex.
Big corporations, including Fortune 50 banks, use ML. However, Alex mentions that startups and mid-sized companies often can’t afford the outlay associated with AI. That’s because it can cost hundreds of thousands or even millions of dollars to build an ML model, implement that model, and cleanse data. If they aren’t using the right tools, it requires a lot of people and capital to make ML successful.
What’ll Happen to Financial Companies That Don’t Leverage ML?
Many companies in finance are still not using ML. “If a bank has $100 billion in assets or less, there’s at least a 50 percent chance they won’t be using any type of AI.”
That provides a problem for companies that haven’t historically been able to afford to use these technologies. They end up absorbing more fraud and identity theft while offering more generic offers to customers. As a result, Alex believes that these institutions will become much less competitive over time than the bigger banks.
What’s the Typical ROI on an ML Project in Finance?
Highly regulated companies usually won’t take on ML projects unless they provide a 100 percent return on investment (ROI) at a minimum. That means these companies are confident of their investment returns every time they invest in the technology.
“Typically, ROI on a machine learning project is between 500 to 1,000 percent,” says Alex. Generally, to produce such lucrative returns, you need to spend tens of millions of dollars on ML technology: “If you couldn’t spend tens of millions, you haven’t had an option to use machine learning in the past.”
Use Cases
Aside from fraud and money laundering detection, ML can personalize credit offers for customers. “If Lisa gets one deal from a provider, I may get another. For example, Lisa may be more inclined to use her credit for cashback benefits, while I may be more inclined to use it for travel,” says Alex. “You can use machine learning to determine the right credit offer to get customers to engage with credit.” ML algorithms can optimize pricing and interest rates to attract customers and encourage the right behavior.
Future Prospects of ML in the Financial Sector
So, what does the future hold for machine learning in finance? Tools and partners like SAVVI AI will enable all companies to use ML, regardless of their size or budget. “Historically, only the biggest banks offered mobile apps,” Alex shared during the podcast. “Now, every bank has one. Machine learning is following the same pattern.”
Takeaways
Machine learning has the potential to save the financial sector hundreds of billions of dollars, while tasks like fraud detection, money laundering detection, personalized credit offers, and custom pricing and interest rates have the potential to revolutionize the industry. Historically, there have been significant barriers to ML for smaller companies that can’t afford this technology, but this is no longer the case.
Thanks to Alex Muller for contributing his insights. If you’d like to hear more from him, listen to the podcast interview we did last year. Contact us to learn how product teams of any size and budget can start to leverage machine learning.
