The topic of artificial intelligence (AI) and its potential impact on the financial industry has gained significant attention in recent times. The Financial Industry Regulatory Authority (FINRA) hosted a Podcast on FINRA Unscripted, discussing the evolving landscape of generative AI and large language models (LLMs).
These new technologies have the potential to dramatically increase efficiency, automate complex tasks, and provide more personalized experiences for customers. However, they also introduce new risks and challenges from a regulatory and compliance perspective that financial institutions need to carefully navigate.
Balancing Innovation and Risk
A significant challenge with large language models (LLMs) is their potential to generate outputs that are biased, inaccurate, inconsistent, misleading, or untrue. These outputs, known as hallucinations, arise when the model is asked a question it does not have sufficient information to answer accurately. Instead of acknowledging its lack of knowledge, the model generates a response that appears plausible but is ultimately incorrect. This can be attributed to gaps in the model’s training data, which may not cover all domains or topics exhaustively, leading to the model making unreliable associations or generalizations when confronted with unfamiliar questions.
Regulations Evolve with Technology
The integration of generative AI systems in the financial industry raises important questions about the adaptation of existing regulations and oversight mechanisms. Regulatory bodies like FINRA and the SEC will need to provide updated guidance and frameworks to address the unique challenges posed by AI-driven technologies. Proactive measures will be crucial to ensure a smooth and compliant transition as these cutting-edge technologies continue to evolve.
In the meantime, FINRA suggests firms look at resources like:
- The 2024 FINRA Annual Regulatory Oversight Report. Specifically, the cybersecurity and the technology management section, which speaks to a lot of what FINRA is going to be asking about in future follow-ups with member firms.
- An executive order released back in late October that’s really focused around the safe and secure and trustworthy development of end use of AI; and
- AI guidance provided by NIST, the National Institute of Standards and Technology.
The podcast emphasizes that close collaboration between financial institutions, technology providers, and regulators will be key to responsibly deploying generative AI in finance while protecting customers. To listen to the full podcast “An Evolving Landscape: Generative AI and Large Language Models in the Financial Industry”, click here.