In February 2022, the AI Public-Private Forum (AIPPF) published its final report (the Report) exploring the pros and cons of the growing use of AI in financial services, as well as how the public and private sectors can drive safe adoption.
Founded by the Bank of England (BoE) and the Financial Conduct Authority (CIF) in October 2020, the AIPPF was created to facilitate discussion between the public sector, the private sector and academia on the safe use of AI. In recognition of the considerable hype around and the complexity of the technology, it also aims to deepen our collective understanding of what we mean by AI and how it works.
What is AI?
As the report admits: AI (and machine learning, a subset of AI) is difficult to define, and as such a broad approach can be beneficial for businesses and regulators. For the purposes of the report, it stops at: “…a branch of computer science, dealing with models and systems for the performance of functions generally associated with human intelligence, such as reasoning and learning”.
AI and financial services
To apply this to a more granular financial services context, as part of the loan origination process, AI could be used to: (i) market the loan, (ii) guide/make the loan decision , and (iii) model the regulatory capital requirement. However, financial services have long used complex mathematical and statistical models to make decisions and provide market insights. As AI can be seen as a complement rather than a replacement for these traditional techniques, we might ask ourselves: what has really changed?
In short, AI can deliver remarkably faster, autonomous, and more accurate models. This comes with a corresponding lack of transparency and explainability, as well as other systematic risks.
Based on Annex I of the report, the following table outlines the benefits and risks of three financial services use cases:
Support safe adoption
Key suggestions on how businesses can support the safe adoption of AI suggested by the report include:
- identify and manage changes to AI models and report on their performance to ensure they behave as expected;
- establish a centralized body within companies to define AI governance standards (made up of individuals with a diversity of skills and perspectives across functions and business units); and
- ensure that there is an appropriate level of understanding and awareness of the benefits and risks of AI across the organization.
It was also proposed that the creation of an industry body for data science practitioners and the professionalization of data science (including voluntary codes of conduct and the establishment of an auditing regime) could foster acceptance and wider confidence in AI systems.
After that ?
With the rapid adoption of AI technology in financial services set to continue, AIPPF encourages businesses and regulators to continue to monitor and support the safe adoption of AI.
The discussions made it clear that the private sector wants regulators to play a role in supporting the safe adoption of AI in UK financial services. As such, the BoE and FCA plan to release a discussion paper after consultation with a broader set of financial services and technology players. The paper will finally look at how AI regulation could work within the existing UK framework (e.g. how it could be applied to senior management and the certification regime) or whether a new approach is needed.
The working document is expected later this year.
Additional contributions were provided by Ellen Blakeney