On Oct. 16, BoE/FCA issued result of survey on machine learning.
In 2019, BoE, FCA carried out joint survey into machine learning in financial services.
Machine learning is the development of models for prediction and pattern recognition, with limited human intervention and is a subcategory of artificial intelligence (AI).
Survey
Asked about nature of deployment of machine learning, the business areas where it is used, maturity of applications and collected information on technical characteristics.
Included how the models were tested and validated, safeguards built into the software.
Types of data and methods used, considerations around benefits, risks, governance.
Key Findings
Machine learning is increasingly being used in financial services in the United Kingdom.
Often, it has passed initial development, and is now in advanced stages of deployment.
Deployment of machine learning is most advanced in banking and insurance sectors.
Machine learning is most commonly used in anti-money laundering and fraud detection as well as in customer-facing applications (e.g. customer services and marketing).
And also in credit risk management, trade pricing, insurance pricing and underwriting.
Regulation is not seen as a barrier but some firms stress need for additional guidance.
Biggest constraints are internal to firms, such as legacy IT systems, data limitations.
Machine learning does not necessarily create new risks but could amplify existing ones.
Firms use variety of safeguards to manage the risks associated with machine learning.
The most common are alert systems and so-called human-in-the-loop mechanisms.
Firms acknowledge that money laundering validation frameworks still need to evolve.
Most machine learning applications are designed, developed in-house but sometimes rely on third-party providers for underlying platforms, infrastructure, such as cloud.
Most users apply existing model risk management frameworks to the applications but these might have to evolve in line with increasing complexity of machine learning.
Next Steps
BoE, FCA to establish a public-private group to explore matters covered in this report.
Jan. 2020 Forum
On Jan. 23, 2020, UK FCA, BoE established a financial services AI public private forum.
Participation in the financial services artificial intelligence public-private forum (AIPPF) is at invitation of FCA, BoE and the selection process is set out in terms of reference.
AIPPF will further constructive dialogue with the public, private sectors to understand use, impact of AI/machine learning, including potential benefits and constraints to deployment, as well as the risks associated with application of AI/machine learning.
Also explore means to support safe adoption within financial services, and whether principles, guidance, regulation, industry good practice could support its safe adoption.
Consider whether ongoing industry input could be useful and what form it could take.
On Oct. 12, 2020, UK PRA and UK FCA launched forum with first meeting, see #88305.