Will AI Truly Transform Financial Services for Everyone?
In a world increasingly shaped by technology, the question arises: can artificial intelligence (AI) truly democratize financial services? On the latest episode of the TechTank Podcast, we delve deep into this critical topic. Produced biweekly by the Brookings Institution’s Center for Technology Innovation, TechTank offers insightful discussions regarding pressing technology issues. Moderators Nicol Turner Lee and Darrell West engage with experts and policymakers, sharing valuable data and innovative policy solutions to address our digitally evolving landscape.
The financial sector has been harnessing AI and machine learning for various applications for a significant time, from streamlining administrative tasks to enhancing decision-making processes. As industries ramp up their investments in these technologies—especially with the recent surge in generative AI—there remains a pressing concern: consumers often lack the clarity on how AI is being leveraged and the potential impacts on their lives.
These complexities are heightened by ongoing innovation efforts within financial firms and the regulatory shifts driven by the previous Trump administration’s goal of ensuring robust American leadership in technology. This has resulted in new societal challenges, including the advent of legislative frameworks that propose regulatory sandboxes for AI companies, coupled with reduced oversight from independent federal agencies.
In this episode of TechTank, co-host Nicol Turner Lee converses with Aaron Klein, a senior fellow specializing in Economic Studies at Brookings and the Miriam K. Carliner Chair. They explore the intricate relationship between AI and the financial sector, discussing consumer implications and the necessity of collaboration in technological advancements.
Listen to the full episode here TechTank Podcast and be sure to subscribe to the TechTank Podcast on your preferred platform: Apple Apple Podcast, Spotify Spotify, or Acast Acast.
CO-HOST NICOL TURNER LEE: Welcome back to Tech Tank, where we dissect pressing technology challenges facing today’s society. I’m Nicol Turner Lee, co-host and Director of the Center for Technology Innovation at Brookings. As headlines increasingly highlight the role of generative AI across various sectors, it’s critical to recognize that certain industries—including finance—have utilized AI and machine learning for quite some time now. From banking operations to fraud detection and credit scoring, AI integration is expansive—and spending on such technologies is projected to rise significantly in the near future. But with this innovation comes a host of policy-related concerns.
Recently, I presented before the House Financial Services Committee, discussing AI’s applications in financial space. Inevitably, while AI holds great promise for enhancing efficiency within the sector, I urged caution regarding its risks—especially given that financial institutions are under strict regulations and responsible for high-stakes decision-making affecting consumers.
After my testimony, I felt it was crucial to delve deeper into this topic with my good friend and fellow expert, Aaron Klein. Aaron specializes in areas including financial technology, regulation, macroeconomics, and infrastructure finance.
GUEST AARON KLEIN: Thanks, Nicol, it’s great to be back with you.
CO-HOST NICOL TURNER LEE: Always enjoy our dialogue. Let’s dive right in. I presented a broad overview last week regarding AI’s applications in finance. Could you provide us with more detail on how AI is actually being implemented in this sector?
GUEST AARON KLEIN: Absolutely, Nicol! Your testimony raised vital points—it's essential to recognize that AI indeed has the potential to address several issues within the current system, but it can also exacerbate existing problems. Instead of getting caught up in binary fears regarding technology, your testimony highlighted a path forward—where we can leverage AI while mitigating negatives.
AI has been an enduring presence in finance; many people are familiar with the FICO score, an established metric used for credit evaluation. The Fair Isaac Corporation—known as FICO—developed this model, which has been a dominant force in credit allocation. However, when it comes to understanding the influence of credit events, uncertainty often complicates things. Many ask, "Will opening a new credit card help or hurt my FICO score?" The unfortunate truth is, even FICO can’t provide a straightforward answer, as it hinges on a multitude of factors—each dependent on others. Therefore, while FICO scores are prevalent, there's considerable ambiguity and potential inaccuracies in this system.
CO-HOST NICOL TURNER LEE: Exactly! While most people know about their FICO scores now, it’s essential to point out that it doesn’t represent everyone's reality. The ramifications of AI further complicate matters, particularly when these systems operate on social patterns or behavioral analytics that may inadvertently discriminate against certain individuals. For discrimination based on racial or gender identification, this could profoundly affect someone’s access to credit.
People might be surprised to learn that data points traditionally used to evaluate creditworthiness include social media activity or even the type of device being used. For example, having a Mac versus a PC might sway your credit scoring decision—doesn’t that seem alarming? In cases where AI may flag someone for not appearing credit-worthy due to their background, it raises immense ethical questions. Moreover, the troubling potential for certain names, often linked to ethnic or racial identities, to be associated with diminished credit eligibility calls for urgent clarification about how these technologies can be managed.
GUEST AARON KLEIN: I completely understand your concerns. While existing credit metrics rest on data from credit reporting agencies—Equifax, TransUnion, Experian—there are significant inaccuracies within these systems. My very own credit report was negatively pigeonholed by someone else’s unpaid phone bill. The ramifications can persist for years, often with very few remedies available for consumers. This black box of FICO can lead to injustices and discrimination. A recent study by Finreg Labs revealed that utilizing cash-flow underwriting—analyzing an individual’s banking activity—might offer better insights into their creditworthiness as opposed to traditional FICO scores.
CO-HOST NICOL TURNER LEE: Absolutely, and this forces us to reflect on how we can incorporate AI in a responsible manner within the financial sector while implementing appropriate safeguards. Clearly, AI shines in areas like fraud detection; it's undeniable that matching algorithms to flag suspicious transactions can lead to significant benefits in our safety. If AI recognizes peculiar spending patterns, like unusual transfers to questionable accounts, it allows companies to target potential fraud proactively.
Yet, the tension between risk-based pricing and non-discrimination laws complicates how we harness these technologies. Here lies a compelling dilemma—on one hand, many believe risk should dictate costs, while on the other, discrimination based on immutable characteristics must be avoided.
For instance, although statistics indicate that teenage boys commonly incur higher car accidents than their female counterparts—which naturally leads to higher insurance premiums—society has deemed it unjust to allow gender as a factor when those statistics lead to inequitable treatment. Thus the subject becomes contentious—how do we establish boundaries on risk indicators, and what happens when AI begins unveiling correlations previously unexamined?
GUEST AARON KLEIN: Exactly, Nicol. AI opens up a labyrinth of correlations between various factors and financial risk. While we might agree that certain behaviors—like texting while driving—should intrinsically count toward increased insurance premiums, we also worry that granular data could lead to wrongful assumptions or unjust penalities.
And do keep in mind the broader societal implications. Legislative measures have emerged to prevent discrimination built upon long-held equity principles, such as those implemented following the Fair Housing Act. The struggle lies in addressing inherent economic disparities while hoping to navigate an opportunity-based landscape created by AI.
Unfortunately, the current political climate is rife with polarization, complicating all discussions regarding AI’s role in finance or any sector. Throughout my tenure at Brookings, I have engaged in many conversations with individuals across ideological lines, seeking to find common ground, yet the current atmosphere dampens these critical discussions.
Additionally, I’d like to highlight some research conducted by Elizabeth Warren, revealing that medical issues and divorce are leading predictors of loan defaults. However, the discomfort surrounding these categories, particularly when assessing them with AI as potential risk factors, illustrates how many factors fuel our lending conversations. Some AI applications can even predict marital status based on financial behavior, an indication of broader societal discussions that many might consider violation of privacy.
CO-HOST NICOL TURNER LEE: Those outcomes hold drastic ramifications. Many of these financial inquiries link back to systemic inequalities. As we have established, AI’s promise to enhance our systems only matters when we rigorously assess their applications to ensure fairness and equity.
It’s essential that we motivate discussions surrounding the prevalence of inequitable practices allowing AI to augment existing disparities rather than alleviate them. Maintaining the dialogue about responsible technology integration into outdated systems is crucial.
GUEST AARON KLEIN: I completely concur. The more we ignore these conversations, the higher the stakes become. Critical issues such as the lack of transparency in AI processes and the integrity of data are paramount, especially when foundational financial systems resist modernization.
Unfortunately, many institutions view innovation and change as mutually exclusive, hesitating to adopt new technologies amid existing challenges instead of seeking opportunities for this inevitable evolution toward success.
This episode concludes with an empty space for future dialogues on how technology can finally serve a more equitable role in financial services, especially amid the challenges we currently confront. Regulatory sandboxes have emerged as a proposed solution where companies can pilot new ideas with certain protections in place. However, the implementation of such initiatives across different states and sectors remains uncertain.
Ultimately, establishing this careful balance between innovation, regulation, and public interest will dictate the effectiveness of AI in driving positive change in finance. To our listeners: What are your thoughts on the potential of AI to democratize financial services? Do you believe the risks outweigh the benefits, or could we find a way to integrate this technology for greater good? We invite you to share your perspectives and join the conversation.
Stay tuned for more episodes of TechTank, where we continue to turn complex issues into accessible ideas. Thank you for joining us!
