The Partnership Between Artificial Intelligence and Humans in Lending

During an era of increased computerization and artificial intelligence, it is often the first instinct of business leaders to automate, replacing humans with computers wherever possible to reduce costs, speed up processes, and minimize error. Algorithms decide so much in our daily lives today – from what ads we see during our online searches to our credit worthiness when making a home purchase. While machine learning can certainly find patterns that may predict mortgage borrower behavior, relying on artificial intelligence alone to make loan decisions is a sure path to a dead end.

At TD, we use AI to help underwriters analyze data (i.e. income calculations, credit scoring, repayment histories based on the down payment amount), but the loan decision is ultimately made by a living, breathing TD colleague. Here's why:

"Computers can crunch the numbers much quicker than your average underwriter, but they cannot provide the contextual analysis that is crucial to the underwriting process"

1.An algorithm alone cannot tell us what we want to know–Are our customers being supported with lending solutions suitable for their circumstances?

Think about the concept of the self-driving car. It can determine the best route for the quickest arrival to our destination, and it reduces driver error that can result in crashes. But, it cannot understand context. A human behind the wheel can see a child on the sidewalk playing with a ball and recognize the potential risk where a self-operating vehicle cannot.

The same is true for underwriting. Computers can crunch the numbers much quicker than your average underwriter, but they cannot provide the contextual analysis that is crucial to the underwriting process. AI helps us make loan decisions faster by eliminating manual input and computation. However, it does not factor in the nuanced relationship between each piece of data. The underwriter understands the life cycle of the individual–where they are in their life and career journey and the potential pitfalls that are multiplied when risks are layered on top of each other. Working together, humans and technology can make faster, fully informed decisions.

2.Often, we hear about "human error," a concept, which in theory, should be eliminated by replacing fallible humans with seemingly infallible computers. But, technology is not exempt from mistakes and biases. 

AI is trained based on the data you feed to it. If the data you are using is biased, so, too, will be the decisions made by the AI. Having a customer-focused colleague who understands the full set of variables beyond the computer's comprehension involved in the process ensures our customers receive fair underwriting decisions and the best possible advice. 

3.And most importantly, there is no replacing human connection.

Buying a home is often the biggest purchase someone will make in their life. The process can be equally exciting and overwhelming. That process shouldn't simply be a computer transaction, but a shared human experience. 

Underwriting is at times the conscience within that experience. When we are loving the ocean view from the deck of our dream home, we may not be thinking about how crippling the high payment will be or how a smaller home could allow us to pay off the loan in 15 years versus 30. 

AI is fast and consistent, but it cannot provide the advice you want and need when making a major purchase like this one. When making your own major life decisions, do you ask for advice from an internet search or a trusted family member or friend? 

Having a dedicated person to answer questions and serve as a guide throughout the home buying process makes this milestone much less stressful for the buyer. That personal touch helps them feel supported, and a quicker, data-driven underwriting decision mitigates risk. Together, the two create a happy customer and a solid loan decision – both of which are good for business. 

As underwriting leaders plan for the future, we must change our thinking from "How can we use AI to replace humans?" to "How can humans and AI work as a team to make quick yet fully in formed decisions and best support the customers we serve?"

At TD, underwriting remains and will remain an equal partnership between the two. We will continue to embrace the technology available to us while still remaining people-powered. After all, we pride ourselves in being Unexpectedly Human.

TD Bank, America's Most Convenient Bank, is a member of TD Bank Group and a subsidiary of The Toronto-Dominion Bank of Toronto, Canada, a top 10 financial services company in North America. The Toronto-Dominion Bank trades on the New York and Toronto stock exchanges under the ticker symbol "TD".