The Federal Reserve is asking for industry comments about the use of AI/ML in banking, as it considers ramping up its oversight of the technologies. The Fed and other stakeholders need to understand the "benefits and risks" of AI as it's incorporated more into financial services, said Fed Gov. Lael Brainard during yesterday's AI Academic Symposium; she specifically mentioned AI's racial biases as a contributor to digital redlining.
- Brainard said understanding risks and benefits is crucial to "ensure that society benefits from the application of AI to financial services."
- The plan is to move forward with a formal request for public feedback about AI adoption in financial services, where banks and other institutions are using it for customer applications, fraud prevention, and operational risk management.
- ML fraud detection tools, in particular, can comb through structured and unstructured data to find suspicious activity faster and more accurately than traditional methods, Brainard said.
- She acknowledged challenges, including the capacity for AI models to lack transparency and have baked-in racial biases. This occurs when historical data with biases are used to train models, leading to bigger "racial gaps in access to credit” and “digital redlining," she said.
- Brainard said it's the "collective responsibility" of the financial industry to build guardrails and protections to prevent these biases and "ensure that AI is designed to promote equitable outcomes."
- Related: Fast Company reported on how AI is quietly causing more institutional racism via high-tech redlining.