Back to General Resources

House Financial Services Committee Task Force Hearing on Artificial Intelligence in Financial Services

House Financial Services Committee Task Force on Artificial Intelligence

“Perspectives on Artificial Intelligence: Where We Are and the Next Frontier in Financial Services”

Wednesday, June 26, 2019

Key Topics & Takeaways

  • Future of AI in Financial Systems: Turner-Lee said that both the private and public sectors are turning to AI systems for a variety of procedures that will disrupt financial systems. Merrill noted that financial institutions will increasingly use machine learning to make more accurate decisions.
  • Future Regulation of AI: Buchanan explained that rapid growth brings challenges regarding regulation, and that AI can fall victim to bias and discrimination. She said AI is not necessarily at fault for bias or discrimination, but the data and programmers are to blame. Merrill stated that regulators already have the authority necessary to balance the risks and benefits of machine learning developments.

Witnesses

Opening Statements

Chairman Bill Foster (D-Ill.)

In his opening statement, Foster noted that financial systems are entering an “exciting time” when it comes to technological advancement, and that artificial intelligence (AI) is at the forefront of these innovations. He pointed out that while AI can be used to efficiently execute trades and prevent fraud, it also provides criminals with more assets to cause harm to the public. Foster stated that it is vital that citizens’ private information is fully protected moving forward. Foster also expressed concern about the continuing practice of corporate consolidation in relation to extremely large data sets.

Ranking Member French Hill (R-Ark.)

In his opening statement, Hill noted that financial innovation can only be accomplished by using AI technology. He stated that AI could benefit cybersecurity, automotive vehicles, the criminal justice system, and the financial services industry. He stated that the goal of the task force is to find ways for AI to enhance compliance obligations, and that AI strategies would enhance institutions’ Bank Secrecy Act (BSA) and anti-money laundering (AML) compliance, as well as for Community Reinvestment Act (CRA) requirements, and that AI has the potential to bring a net positive gain of jobs to the economy.

Testimony

Dr. Nicol Turner-Lee, Fellow, Governance Studies, Center for Technology Innovation, Brookings Institution

In her testimony, Turner-Lee noted that the private and public sectors are increasingly turning to AI systems and machine learning algorithms to automate simple and complex decision-making processes. She said that AI will disrupt most economic sectors including transportation, retail, advertising, financial services and energy. Turner-Lee noted that AI models are starting to reveal troubling examples in which the reality of algorithmic decision-making falls short of expectations. She expressed concern about the potential risk of bias that AI will bring against Latinos and African Americans, saying that Congress must modernize civil rights laws and other consumer protections to safeguard protected classes from online discrimination.

Dr. Bonnie Buchanan, Head of School of Finance and Accounting and Professor of Finance, Surrey Business School, University of Surrey

In her testimony, Buchanan noted that AI offers the possibility of increased financial inclusion, but rapid growth brings challenges regarding regulation. She said that AI could introduce bias and discrimination and that deep learning techniques “provide predictions, but they do not provide insight into how the variables are being used to reach these predictions.” She stated that policymakers need to be concerned with the explainability and accountability of AI models. Buchanan pointed out that black box modeling must be completely avoided.

Dr. Douglas Merrill, Founder and CEO, ZestFinance

In his testimony, Merrill noted that because machine learning-powered credit scores outperformed traditional credit scores, companies will increasingly use machine learning to make more accurate decisions. He noted that customers using their company’s machine learning underwriting tools have seen a “15% approval rate increase for auto loans, and a 51% increase in approval rates for personal loans — each with no increase in defaults.” A problem that Merrill noted is that when it comes to AI, many explainability techniques are inconsistent, inaccurate, computationally expensive, or fail to spot discriminatory outcomes. Merrill stated that regulators already have the authority necessary to balance the risks and benefits of machine learning developments. He stated that as Congress continues to deliberate on how to implement AI, tens of millions of Americans will be excluded from the credit market or poorly treated by it.

Jesse McWaters, Financial Innovation Lead, World Economic Forum

In his testimony, McWaters noted that AI is transforming the operating models of financial institutions because of the accessibility and personalized accessory capabilities. He said AI may redraw the map of the financial sector, and smaller financial institutions may have to turn to third party AI services to keep up in the competitive landscape. He said the use of broader data sets offers risk to consumer privacy and biases, and though McWaters acknowledged the risk that AI brings, he explained the risks must be put in context with opportunities that come with AI. He said AI allows for more accessible, high-quality financial advice, which can help smaller community banks become more digitally relevant to consumers.

Question & Answer

Future of AI in Financial Systems

Rep. Barry Loudermilk (R-Ga.) asked why few financial institutions are using AI for fraud prevention. Buchanan explained that fraud is not directly observable, which is difficult for machine learning technology to detect.

Rep. Anthony Gonzalez (R-Ohio) asked about balancing innovation with protecting consumer data. Buchanan clarified that “big data” is not the same as “strong, robust data,” adding that AI utilizes “strong, robust data” that is more concerned with privacy.

Foster asked if there are any agreed upon metrics for measuring fairness in AI. Merrill responded that machine learning is biased because many of the programmers were white males, and that Creation Upfront Requirements would prove to be “extremely” helpful.

Regulation of AI

Hill asked how to encourage regulators to update the 2011 rulemaking regarding machine learning. Merrill explained that regulators are not against innovation. Hill followed up on his question by expressing his support for the uniformity of the “sandbox” idea in which machine learning products can be tested in a closed environment before going the public.

Foster asked about the extent to which companies should be required to audit algorithms and artificial intelligence so they don’t unfairly discriminate. Turner-Lee explained that instituting of auditing practices are needed to ensure there are no unintended racial or economic bias. Turner-Lee further explained that it is crucial that there be more discussion of how algorithms would comply with nondiscrimination laws prior to their launch in the public domain.

Rep. Alma Adams (D-N.C.) asked about safeguards and regulations Congress could put in place to ensure AI systems do not have a devastating impact on vulnerable communities and consumers. Turner-Lee explained that the developers of AI algorithms are not representative of the population who use them, and suggested that in order to mitigate this issue, there needs to be more diversity among those who create the algorithms. Adams asked if it would be useful for Congress to fund algorithmic bias research through the National Institute of Standards and Technology (NIST) for federal agencies to develop tools to resolve bias, to which all witnesses agreed.

Rep. Ted Budd (R-N.C.) asked about the regulation of AI and if overburdensome regulations stunt growth. McWaters explained that the perspectives of foreign countries on U.S regulations is very complex, which deters innovation.

For more information on this hearing, please click here.