5 minutes
With a disciplined approach, CUs can successfully leverage this technology today.
In the last two years, most technology companies have started using the terms machine learning and artificial intelligence. In fact, there has been a 1,400 percent increase in the mention of these two terms during earnings calls of public companies during the same period.
Artificial intelligence in its essence is analogous to an army of statisticians building predictive models or algorithms in order to make decisions or recommendations. Machine learning techniques could in theory create this artificial intelligence without human intervention or, more realistically, with fewer humans. In this discussion, we are going refer to machine learning-powered artificial intelligence as intelligent algorithms.
Financial institutions can already leverage the power of intelligent algorithms. There is no need to wait for some magical and massive scientific breakthrough. Success in this pursuit is directly correlated to hypothesis-driven methodology and use cases. As boring as “hypothesis” and “use cases” may sound, they drive success at the firms that have successfully implemented such algorithms.
What is 'hypothesis-driven methodology' and why it is important?
The purpose of intelligent algorithms is to make decisions in the midst of uncertainty with as few errors as possible. Hypothesis-driven methodology defines the purpose of these algorithms and creates the boundaries within which the intelligent algorithm would operate before needing to hand over control to humans.
Three pitfalls that this hypothesis-driven methodology helps avoid are:
1. Road to Nowhere: All algorithms are purpose-built. If a financial institution needs a portfolio loss prediction model for commercial loans, the predictive model would need a clear purpose, i.e., a hypothesis about what data would be needed to make such predictions, what outcomes would be considered statistically significant and what modeling techniques would be relevant and meaningful for this algorithm. Without clear methodology, it is easy to end up on the road to nowhere.
2. Black Box of Magic: Several vendors tout the prowess of their tools, which can somehow magically garner insights from any industry-specific data source. While this approach may be successful to mine such standard information as social media feeds, a “black box” approach that is not custom-built for industry-specific and business model-specific needs is doomed to fail.
3. No Data Scientists Needed: If a financial institution wants to pursue intelligent algorithms but has no one on the team who understands concepts such as linear regression, correlation or statistical significance, the odds are quite low for success. The notion that somehow machine learning techniques can magically usher in artificial intelligence at a credit union without the need for human data scientist resources is folly. Data analysts and data scientists are needed to create the hypothesis—the purpose for these algorithms—and then validate, back-test and calibrate these algorithms in order to drive the success of such initiatives.
Use Cases That Can Be Leveraged by Credit Unions Today
Across various industries, many intelligent algorithms are being successfully used to make recommendations, forecast the future, find insights and reduce operating costs. Credit unions can leverage such specific deployments to achieve the same success.
Here are four use cases to consider:
1. Predictive Models: Firms such as CapitalOne have been successfully leveraging predictive models since the stone age of artificial intelligence, the 1990s. Financial institutions are rich with data from core, online, card and credit bureau platforms. Through a clear, hypothesis-driven methodology as described above, a financial institution can create and implement its own predictive models with off-the-shelf tools available in the market, especially on its card portfolio where the profitability and transaction volumes are rich.
2. Prescriptive Models: At just about all banking-related conferences, it is close to impossible to escape the buzz phrase next best action, which at its core is a prescriptive model-driven outcome. “Next best action” is simply an integrative thought process of recent events in order to recommend the next conversation, advertisement or action. Processing power to create these integrative thinking algorithms, either as an in-house deployment or through a third party, is very affordable. A clear set of business rules and accumulation of data will determine the next best action to promote product offerings and reduce attrition.
3. Voice Mining: An often-overlooked area of analysis that has been very mature since the mid-2000s is voice or speech mining. The English language only has 44 unique sounds, called phonemes. Intelligent algorithms can review voice recordings and mine these phonemes in real time to provide the financial institution with in-depth information on why a customer called, identify patterns and help resolve root causes of issues in order to reduce operating costs. Walking around credit union’s call center and asking the staff, “What is going on today?” can become an opportunity to tend to employees rather than a daily data-gathering necessity by implementing these highly intelligent and highly mature algorithms.
4. Text Analysis: The same phoneme-based algorithms can intelligently query a vast amount of free-form text information and provide in-depth insights from a financial institution’s copious amounts of data. This same technology is behind several automated chatbots that respond to email and text-based customer inquiries and subsequently reduce operating costs for many organizations.
In summary, all financial institutions can get out there today, follow a disciplined hypothesis-driven methodology and leverage existing technology to create and deploy purpose-built intelligent algorithms through clearly defined use cases. There is no need to wait for some amazing artificial intelligence technology to arrive, because it is already here!
Niel Devasir is senior director at CUES Supplier member and strategic partner Cornerstone Advisors, Scottsdale, Ariz. With 20 years of financial services experience, he has a track record of building high growth and operationally efficient companies by leading mergers and acquisitions and overseeing operations, technology, corporate procurement and analytics functions.
