Artificial Intelligence in Financial Services

The FCA is consulting with firms regarding how artificial intelligence (AI) may affect financial services. This follows the publishing of the final report on the AI Public-Private Forum (AIPPF) by the Bank of England and the FCA in February. The report aimed to further dialogue on the innovation and safe adoption of AI and Machine Learning (ML), exploring the various barriers and challenges in relation to AI, as well as potential ways to address them.

With the new consultation, the FCA wants to gain feedback on:

• the potential benefits and risks related to the use of AI in financial services
• how the current regulatory framework applies to AI
• whether additional clarification of existing regulation may be helpful
• how policy can best support safe and responsible AI adoption

Submissions for the FCA consultation close on 10 February 2023.

AI vs. ML: what’s the difference?

Artificial intelligence is the capability of a computer system to mimic human interaction in order to independently carry out functions such as learning and problem-solving.

Machine Learning is considered a subset of AI, and although the two are closely connected in their processes, they are technically not the same thing. ML is how the computer system develops its intelligence, using mathematical problems and data to improve its own learning without requiring instructions to do so.

Essentially, ML allows a computer to learn, AI allows it to think.

How would businesses benefit from AI?

The use of AI or ML in financial services can enable firms to offer better products and services, which in turn could lead to better outcomes for consumers, firms, financial markets, and the wider economy. Implementation could also potentially improve operational efficiency, boost revenue, as well as drive innovation within the insurance sector, in relation to how risk data is calculated.

One application of AI is to process customer data or respond to customer queries in real-time through AI enabled ‘chatbots’, improving the customer experience by offering faster query resolutions compared to conventional – and sometimes slower- customer support journeys. This may include more effective matching to products and services, enhanced abilities to identify and support consumers with characteristics of vulnerability, as well as increasing financial access.

Application of AI within Insurance

Despite the benefits of improved services and the value of data collection, the insurance sector is still reportedly falling “behind the curve” compared to other sectors in its adoption of AI and MI, particularly when it comes to manual processes and document processing. As Sameer Deshpande, head of enterprise architecture at Broke PIB Group explains, “the way we interact with our customers is really a low-point – it’s still very dependent on central operations.”

If the insurance sector wishes to expand its use of AI over the next few years, then it would need to move away from central operations and call centres, so that it can catch up with the levels of service other industries are already providing to their customers within financial services.

Considerations and Guidance

As is the case with most innovations within the industry, it is important for firms not to rush headlong into adopting AI practices without first considering the pros and cons. The use of AI and MI is still relatively new, and the promise of efficiency and accessibility is an enticing lure. However, it still comes with the caveat in that it can create significantly more problems or amplify existing ones if not implemented correctly.

For example, if AI data is misused, it could potentially lead to harmful targeting of consumers’ behavioural biases based on their characteristics of vulnerability, which could result in discriminatory decisions, financial exclusion, and ultimately, reduced trust.

AI relies heavily on large quantities of data, which means there is also the greater risk of error in its processing and storage, resulting in data becoming incomplete or unrepresentative if left unmanaged. Poor management of data may potentially leave firms open to a greater risk of cyber threats if the software is unable to detect malicious content within its systems.

The FCA, Bank of England, and Prudential Regulation Authority have stated that they have “a close interest in the safe and responsible adoption of AI in financial services” in line with their statutory objectives. This may involve intervening further to mitigate any potential risks and harms related to AI applications, including considerations for how policies and regulations can best be supported.

To submit feedback or read the full discussion paper from the FCA and the Bank of England, click here.