top of page

AI, Mortgage Banking, and a Lot of Mixed Metaphors

by Ruth Lee, CMB

Picture this: it's a Tuesday afternoon, and the OpenAI CEO, Sam Altman, is navigating the hot seat, fielding questions from the Senate Judiciary Committee about Artificial Intelligence (AI). The discourse? The future of AI and the urgent need to define its boundaries before we find ourselves in an episode of Black Mirror.

In a move that startled many, Altman implored Congress to regulate his industry, likening AI to a bull in a china shop during his testimony. According to Altman, artificial intelligence (AI) is not just knocking on the china cabinet; it's already rummaging around in the porcelain. Senators raised their eyebrows at this appeal. Sen. Richard Blumenthal even quipped about the potential for AI being more like a bomb than a bull in the china shop. And I love mixing metaphors - so Bravo! Sen Blumenthal, Bravo! But, I'm just sayin' when even the bull asks for a leash; it might be time to consider getting one.

Altman's prophetic plea? "Regulate us, pretty please, before we accidentally break the world." (Okay, so he didn't say it quite like that, but I'm taking some poetic license here). As we delve into the heart of this plea, we find ourselves staring at the elephant in the room - the concept of 'garbage in, garbage out' in AI and how it weaves into the fair housing narrative. Brief pause: for those of you who've been in the mortgage industry for over a few weeks, who among us doesn't live and eat by the adage of "garbage in - garbage out?" So, yes, I am paying attention.

I've been doing a lot of research on artificial intelligence and machine learning. I'm learning Python and taking math for the first time in thirty years. I want to differentiate its highest and best use and where it veers into the realm of the meme coin for mortgage banking. As an industry, we love a shiny ball. What wasn't hard to figure out? Serious AI is a data-devouring beast. Feed it biased, unrepresentative, or just plain garbage data, and it'll spit out skewed, unjust, or garbage results. This is particularly crucial in areas like mortgage banking, where historical discriminatory practices such as redlining have already tainted some of the data pool. This isn't insurmountable - it just has to be acknowledged and addressed.

AI and machine learning models are powerful tools that can significantly improve efficiency and decision-making in mortgage banking. They can be used to automate underwriting processes, predict loan defaults, identify fraud, and provide unprecedented risk mitigation across the entire market. However, these models are data-driven, and their effectiveness and fairness largely depend on the data quality they are trained on.

If we are not careful, our shiny new AI systems could perpetuate the discrimination and inefficiencies we've tried to eliminate for decades. And let me preach to the choir - FAIR HOUSING IS GOOD FOR BUSINESS. Plain and simple. If you deserve to be a homeowner, let's get you a loan. And if you don't? Well, let's find out early and stop wasting time. So here are some new terms that will be as fun as "leakage," "synergy," and "disruption" in mortgage banking - welcome "Bias-Aware Data Collection," "Fairness-Aware Machine Learning," and "Continuous Monitoring" to ensure that our AI systems are working for everyone, not just a privileged few.

Sen. Blumenthal reminded us that we've been here before. Remember the 'Wild West' days of social media? When potential risks were overlooked in favor of rapid innovation and growth? We're still dealing with the fallout from that one. So here we have both senators AND the tech industry determined to avoid a repeat of that fiasco. Altman expressed this exact fear of the potential abuses in AI as the reason he started OpenAI.

Now, this is where the ideas get interesting. Gary Marcus, NYU Professor, suggested the idea of "AI nutrition labels." Just like the labels on food that tell us what we're putting into our bodies, these would inform us about the data diet of our AI systems. It's a powerful concept, as it could provide a level of transparency that's often missing in AI. Altman proposed a three-pronged approach to AI regulation. Establishing a federal agency to oversee AI development, enforcing safety standards for AI models, and conducting independent audits. None of these things are new for the highly regulated mortgage industry.

So, here we are, standing in a china shop with a bull called AI. It's powerful, potentially reckless, and difficult to control. But, as Altman said, it can also be a "printing press moment." We've got the chance to harness this technology and use it to create a more equitable society. But to do that, we must work together to ensure that our AI systems are fair, transparent, and regulated. I am all in and completely fascinated. Please like, comment, or follow me for what will be a fascinating ride with AI and its impact on my favorite industry. No trolls, please.

36 views0 comments


Rated 0 out of 5 stars.
No ratings yet

Add a rating
bottom of page