LendingClub: AI Can Transform Lending and Broaden Inclusion

The artificial intelligence (AI) revolution has been more of an evolution.

“AI is the latest buzzword for something that’s been building for at least a decade,” Anuj Nayar, financial health officer of LendingClub, told PYMNTS, “and that’s the use of technology in financial services.”

The headlines may make it seem like AI has suddenly arrived to shake up the world. But in fact, there’s been a long-standing effort to use tech in order to streamline and update back-end processes, while cutting down on manual tasks.

The modern age of tech, he said, can be traced back through the past few decades, when traders began using algorithms to buy and sell commodities and securities. And in the present day, lenders are using some of those same features and principles to extend loans and other offerings across platforms in automated fashion.

But those activities are becoming table stakes.

Now, Nayar said, there’s the opportunity to employ AI and machine learning — in tandem with automation — to use data to include people in the traditional financial services ecosystem who might have long been excluded.

“That’s the Holy Grail,” he said.

The traditional methods of determining creditworthiness, have, of course, rested on the FICO score. As Nayar put it: “The score has five attributes, and the score is what you were given.” The advent of the platform model, and in LendingClub’s case, the ability to collect and analyze billions of data points — and thousands of attributes — makes it possible to get a much more accurate assessment of a would-be borrower’s profile.

Avoiding Unintended Consequences

But in order to harness that data, he said, “you have to use technology and machine learning to make this all happen.”

He noted, too, that, even after the loan’s been extended, there’s the benefit of using AI to better manage the customer-facing processes throughout the loan lifecycle.

But there’s some caution in the mix, Nayar said. As with any new technology, it’s critical to ensure that users — the individuals and the institutions — are careful to avoid unintended consequences. That means meeting fair lending guidelines and ensuring biases don’t creep into the models and the algorithms themselves. The key is to avoid a “black box” effect where users are unable to “break down” the models to explain the decision making — and why some applicants got loans, and not others.

Nayar noted that forward-thinking companies, LendingClub among them, have tech, audit and compliance teams in place to ensure vigilance against that bias. With a nod toward his firm’s own efforts, he said, there’s the feedback that’s given to applicants who were denied as to just why they’ve been turned down — and consumers can contest the decision if the data proves incorrect.

As he told PYMNTS: “We don’t underwrite based on ZIP codes or who someone’s friends are based on who their friends are on social media.”

We’re right now in the third iteration of the web. We were all on our desktops 20 years ago. The idea that we could bank online as opposed to visiting a branch seemed scary back then. Then came banking via mobile phones, of course. And now, he said, technology has become so pervasive that it’s becoming easier, and expected, that financial services would be embedded into everyday life and online/offline activities.

“It’s still early days,” Nayar said, “and we’re in an interesting time for the use of technology … and we’ll see a fundamental shift in financial services.”