Can AI move the dial for the consumers of financial services and products?

The FCA has just released a research note on the subject of how useful generative AI might be in influencing change in the consumer-facing space in financial services. The focus of the FCA's research project was two-pronged:

• Can generative AI simplify ‘complex financial concepts’ enough for those consumers who have lower levels of literacy and numeracy?
• Can chatbots effectively deliver Q&A-style guidance about cash savings products?

Test and trust

The FCA makes some important headline observations about the regulatory approach and about their expectations of firms looking to deploy generative AI solutions:

• the regulatory approach is outcomes-focused;
• testing is a key component to delivering improved consumer outcomes;
• testing will reveal the issues that need to be solved in order for AI to have a positive impact on consumer outcomes; and
• positive interactions with consumers, better consumer outcomes, and improved consumer trust are interlinked concepts. 

Why test AI's ability to assist financial decision-making?

We know that there are many challenges for consumers in finding their way around financial services. There are many products for them to choose from. There are many different sources of information for them to access as part of their decision-making process. There are many complex concepts and computations for them to take on-board. These challenges can lead to consumers placing financial decision-making in the too hard box. Could AI help here to facilitate the delivery of easier consumer journeys and better consumer outcomes? 

Lessons learned

Although this research had a narrow focus, the lessons learned from it may have a more general application and demonstrate that:

• generative AI seems able to ‘accurately simplify concepts and make text easier to read’;
• testing the output of generative AI requires both human and non-human evaluation (with testing by real consumers as the ultimate measure of how readable and understandable something is);
• how and where generative AI is used in the consumer journey will likely impact the consumer outcomes; and
• consumers are receptive to some forms of automated support. 

Dig deep

Detailed testing and evaluation is needed to check for a range of possible issues with generative AI output, including quality, accuracy, misinformation, jargon, and appropriateness (including of imagery).  The results of the FCA's research projects indicate that ‘seemingly minor changes can influence customer understanding and product selection’. Given that the research scope was limited, the research results would seem to underscore the need for firms undertaking their own testing to fully explore the effects of, for example, language and imagery, in the context of the capabilities of the specific generative AI that they are using, and the specific questions that they are trying to answer, and inform their own learning accordingly. 

Consumer satisfaction survey

By asking consumers who take part in their research surveys additional questions, the FCA are able to collate evidence about where and how consumers might be willing to engage with AI, including chatbots. The results indicate that consumers will engage with AI to:

• make comparisons between products;
• simplify jargon;
• solve queries;
• seek advice; and
• source the information that is needed in order to make financial decisions. 

The way forward

The FCA is committed to pilots and testing, and to working with firms to explore questions and evolve learning, to explore the opportunities that AI can bring to financial services and to the consumer journey.  

Regulatory focus

If you would like to discuss how current or future regulations impact what you do with AI, please contact me, Tom Whittaker, or Martin Cook. You can meet our financial services experts here and our technology experts here.

You can read more thought-leadership like this by subscribing to our monthly financial services regulation update by clicking here and clicking here for our AI blog and here for our AI newsletter. 

Our lessons show that there are important opportunities to provide effective and tailored support with AI. For example, this could include providing financial guidance, tailored explanations of financial concepts, or agent-based chatbots for resolving complex consumer queries. However, they also show that careful testing is important and that we still have lots to learn about how to use AI with consumers.

https://www.fca.org.uk/publication/research-notes/research-note-lessons-2-large-language-models-pilots-consumer-guidance.pdf