Generative AI and machine learning are the new ‘sexy’ across just about every industry, and fintech is no different.
Riding this wave of industry disruption are software companies promising banks and lenders more autonomy with user-friendly solutions. Some include drag-and-drop models for real-time underwriting and risk decisioning.
No coding necessary. No maintenance needed. Just about all your decisioning on autopilot.
And unfortunately, lenders are falling for it.
Drag-and-drop AI-driven models represent a very appealing offer. Lenders could potentially save on human capital by automating their decision engines, essentially saying ‘goodbye’ to expensive data science teams.
The problem is: these models, left on autopilot, have the potential to miss very important considerations.
To remain competitive in this space, it’s almost mandatory to have top-of-the-line Gen AI and ML models working in the background.
But here are a few reasons why lenders still need humans – teams of very smart people knowledgeable about data science, financial services, regulations, and tech – managing these systems.
Federal Consumer Protection Laws
Number one on our list are federal consumer protection laws.
Most lenders work within the confines of over 18 different federal laws, including the Truth in Lending Act (TILA), Equal Credit Opportunity Act (ECOA), the Electronic Fund Transfer Act, the Graham-Leach-Bliley Act, and more.
Some are in place to make sure consumers are treated fairly without bias. Others govern how banks and lenders manage the risks associated with third-party relationships.
A fundamental knowledge of existing regulations, such as OCC 2011-2012 and FIL-22-2017, and any newer ones that may arise are critical for model building.
Millions of dollars could be at stake if a lender ends up on the wrong side of a federal audit without a way to defend the elements of their decision engines. And “We didn’t know because we let AI automate it for us” simply isn’t an excuse.
People and Market Changes
Student loan repayments. Inflation. Housing market conditions. Government incentives. Changes in currency. The circumstances that impact people’s motivations to seek out financing and their ability to repay evolve quickly.
Current Gen AI and ML tools are good at considering historical data or all the factors we’ve seen before at lightning speed. But what about the emerging factors that we haven’t seen before? Again, you need teams of people daily monitoring these macroeconomic evolutions and how they could potentially impact your model development.
Other Emerging Problems
Robust decisioning models must also adapt to emerging problems.
For example, Bloom Analytics has developed a proprietary model to score leads, ranking them by quality or likelihood to pass underwriting parameters. Sometimes, companies profiting from selling leads adapt and attempt to funnel leads that aren’t the best fit.
However, the Bloom model is so robust that it can respond to any deceptive practices to funnel bad leads that may surface.
But that’s because there’s a very knowledgeable team behind the models who are paying attention to emerging problems and making updates accordingly.
Wrapping It Up
Don’t get us wrong. The objective here isn’t to paint AI decisioning tools as completely useless. In fact, they are quite effective as productivity tools – but for teams that know what they’re doing.
The danger of drag-and-drop, fully AI-driven solutions is that, by their very nature, they hide the technical boring details that most would rather not be bothered with. You give up ownership and the power to fully customize your models. And the devil is in the details when it comes to model building.
AI has not made humans obsolete.
An educated team is the key to success in this ever-adapting space. So, please, please don’t throw your hands in air and leave everything to drag-and-drop.
Humans are still very relevant in fintech.
Learn more about how Bloom Analytics employs AI to optimize lead flow management, lead quality, risk evaluation, underwriting and more. Schedule a consultation today.