Accountability: the FCA on Artificial Intelligence in Financial Services

Christopher Woolard, Executive Director of Strategy and Competition at the FCA, gave an interesting speech at the AI ethics in the financial sector conference at The Alan Turing Institute yesterday. As Flexys conducts its own artificial intelligence research and development exclusively for the debt collection market, we were listening intently. From the regulator’s perspective, there are three aspects to consider: continuity, public value and collaboration.

The speech promoted the importance of balancing innovation in artificial intelligence with the interests of consumers and social responsibility. The benefit to customers must be the driving force behind AI in the debt collection sector because without a fair exchange it will be hard to convince customers to share data and participate. Transparency, from AI application providers to debt collection operations and between businesses and their customers is essential and should be embraced and transmitted to the public enthusiastically. 

Woolard also noted that businesses deploying AI must know the provenance of the technology they are using. This is much easier if organisations are in direct contact with the creators of the algorithms who can then authoritatively answer any questions asked about function, ethics, explainability and security. It’s easy to see how black box AI applications designed and created by unnamed people whom businesses will never meet and cannot question, may cause company boards to baulk at the risk of the unknown. If, as Woolard envisages, boards ask ‘what’s the worst that can possibly go wrong?’ a competent AI development team will have the answers- they will have already asked themselves those questions throughout the R&D process.

In the debt collection environment, businesses are highly likely to require some degree of explainability, especially when customers may be experiencing vulnerable circumstances or where debt collection activity may be questioned in the public arena. As Woolard explains, collaboration across industry, regulators, academics and the public will be necessary. To this end, Flexys have secured a Knowledge Transfer Partnership with The University of the West of England Artificial Intelligence department to build explainability into our AI engine and to draw on the complementary skills of the UWE academics.

The FCA naturally wants customer-centricity at the heart of AI development. Any trade-off in terms of access to and use of data must keep the customers best interests at heart. Organisations can still benefit considerably from efficient and effective strategisation due to AI. So, customers in debt who engage with an AI-enhanced collections process can be reassured that their data will be used to personalise the journey they encounter and to direct them quickly to the most appropriate action. They can also be the beneficiaries of the increased context-awareness that AI can bring to digital collections, with the aim of a more accessible and responsive customer engagement and service capability.

At Flexys, collaboration and transparency are the founding principles that govern all our AI research and development and because it’s in-house, we have full control and accountability over our machine learning algorithms. We listen to collections professionals and specialists in vulnerability and mental health to create secure and ethical artificial intelligence modules that increase efficiency, drive better outcomes and improve customer experience.