Tensions surrounding the use of artificial intelligence continue in the financial services industry. Ann Implications for economists The report found that while four in five industry executives agree that the use of AI will separate winners from losers, nearly two-thirds also believe that the complexity and risks often outweigh the benefits. In fact, because financial institutions are highly regulated, the risks in how data is used and technology is applied to customer-facing services are higher than those encountered by organizations in other industries.1 However, secure AI innovation is not impossible for financial services, and there are several use cases where these organizations can take advantage of new technology without exacerbating risk.
Many of the practical applications of AI in financial services are simply further evolutions of existing uses of AI. Some early use cases that can be implemented in a controlled, low-risk manner while still having a high impact include:
- Disputes and Investigations. In discovery and document review, legal teams are increasingly testing the effectiveness of various generative AI tools. Using samples of clean, secure data allows organizations to test language models at scale without exposing themselves to unnecessary risks or raising ethical concerns. One approach is to apply generative AI to review documents related to previous issues and compare the results to human review decisions. This provides a baseline of how effectively AI performed against existing methodologies. Generative AI tools can also proactively identify privilege based on parameters set by legal teams, helping automate the logging of privileged content.
- Reupholstery practice. Optical character recognition can significantly speed up the process of analyzing handwritten documents and low-resolution scans for a variety of problems, including tedious paper reprinting tasks. By automating most manual steps in contract review and re-contracting, legal teams can reduce the time and cost required to complete the task without sacrificing quality or accuracy, especially if generative AI proves to be highly reliable.
- compliance. Generative AI models can be designed to support summarization and reporting across structured databases for compliance purposes. For example, a model can ingest large amounts of structured data, such as a series of transactions or financial accounts, and summarize and report trends and anomalies in the data. If compliance teams ensure that accurate, high-quality data is entered into models, these tools can accelerate time to insight without negatively impacting sensitive information. Additionally, generative AI tools can automate voice-to-text cleanup and review by removing errors and filling in gaps from transcriptions of recordings, making it easier to review and analyze for compliance purposes.
- Data Subject Access Requests. Large-scale language models have shown significant results in their ability to identify and extract personal information from large datasets. For financial institutions that need to comply with various privacy laws around the world, this makes it much easier to find and create information, including certain personally identifiable information, and reduces the time and cost of responding to data subject access requests.
Financial institutions often face much higher risks than organizations in other industries, so it’s important to balance risk management with AI innovation. Global AI regulation is also evolving, with EU AI law serving as a benchmark for the types of requirements that authorities will enforce regarding the use of AI.2 Explainability is one of the key areas, and legal teams should proactively review existing and future uses of AI to ensure they meet explainability standards and that the models being deployed do not violate privacy requirements, financial reporting regulations, or other legislation.
Legal and technology experts can partner to establish parameters for how AI is used in legal and compliance environments and guide governance around the adoption of new technology in other parts of the business. Combining technical and legal thinking is critical to helping financial services institutions achieve their AI goals without introducing excessive risk.
