Financial services companies have been using artificial intelligence in a responsible, regulated way for years to better serve their customers. In this edition of Fintech Explained, Zest AI CEO Mike de Vere unpacks how companies like Zest AI use artificial intelligence to expand access to credit and how Washington can approach AI policy moving forward. Dive in to learn more.
Q: Let’s start with the basics: Tell us about Zest AI and its mission.
As we all know, the current credit system is crushing vast segments of our nation and their ability to make better lives for themselves and their families. Zest AI’s technology is solving this huge problem by automating consumer credit underwriting with more accurate and equitable lending insights — supporting our mission to broaden access to credit and enable financial equity for all.
All financial institutions need tools that will help them make smarter, faster, and more equitable lending decisions. They also need ways to monitor the health of their business so they can continue to lend in the future, which is why we were incredibly excited to announce our new Zest Portfolio Management product last month.
Q: Can you explain how AI – and AI developed by Zest AI in particular – expands access to fair credit for millions of Americans?
Humans are important to lending, but bias does exist in our current credit system. For example, an underwriter’s morning coffee might not have kicked in for their application, or they could be coming back from a big lunch or facing some relationship issues. The fact of the matter is that humans are inconsistent. But things that distract a human can’t distract an equitably built algorithm.
Humans using resources like better math through machine learning makes it so we don’t miss risk indicators and don’t add them in where they don’t belong. That means folks overlooked by old ways of determining creditworthiness because of their race, ethnicity, gender, or other factors are getting a fair shot at credit. Our algorithms, with better data and better math, are removing the biases.
Q: In your view, what are the hallmarks of a company using AI responsibly and ethically?
There are a couple of shining star hallmarks when it comes to responsible AI. And it goes beyond AI being fair and transparent. Here are some key elements to that:
- The AI’s algorithm is built to find the most equitable decision — it goes above and beyond the current bar for fairness.
- The AI has a diverse group of people building it and then monitoring it. Responsibly built AI will keep its algorithm simple when it can stay simple — complex models should be reserved for when they’re necessary, not when they look impressive.
Q: There is currently a debate in Washington and on Capitol Hill about taking a targeted vs. a comprehensive approach to AI legislation. Which approach do you think is best and why?
Interestingly, we’ve already laid a lot of the foundation for AI legislation in the financial services industry. Compliant AI is monitored, requires fair lending testing, and must follow the rules that the regulators have set in place for the humans making lending decisions — which it’s capable of doing.
What we need to do now, both in the lending industry and throughout other industries, is define and regulate the specific cases that pop up as AI innovations advance. We don’t want to stifle creativity, but we want to be purposeful about protecting end consumers from any harm or discrimination that crops up when AI is used for selfish purposes.
AI can be built to promote equity consistently and transparently. Protective, targeted legislation can help ensure that those algorithms are accessible to all.
Q: Let’s end with one fun thing: Name a use case for AI you’re most excited about in the coming years.
Finding ways that AI can help enhance our healthcare system is an exciting prospect. The human experience is super complex. It would be incredible if AI could apply the plethora of health data to personalize treatment plans to help someone heal and get better faster. And since we already know how to remove bias from lending through AI, being able to apply that practice to medicine would be a huge win for medical equity.