The Financial Conduct Authority said data will be the ‘blood flow’ to the organisation and will help staff hone in on their expertise, as part of its focus to become a data-led regulator.
Speaking at the Association of British Insurers’ annual conference today (February 22), Ian Phoenix, director of digital and intelligence at the FCA, discussed how the regulator will use data to maximise its value.
He said: “What does it really mean in practice? It means making optimal use of our limited resources so our people spend more time on the things that require expertise. It means making sure we don't grant challenges for too long.”
Phoenix explained that in order to deliver this, it is important to maximise the value that individuals already have, making it more useful and providing the right tools for frontline staff to help them make the right decisions.
“We also want to work more effectively with many of our partners to ensure we can quickly identify insights from data to determine the best outcomes for the consumer market,” he said.
“There are ethical challenges with data and we need to find the right balance to maximise outcomes in compliance. We can continue to make greater use of information that's online, including social media, to adapt to a changing world where financial promotions appear in an Instagram ad just as much as it can in a newspaper.
“When we get this right, we'll be able to use this data as our blood flow to the organisation, feeding information to the right place at the right time.”
Data science vs AI
Phoenix also discussed the difference between data science and artificial intelligence and how the regulator will use this.
As part of the FCA’s ambition to be a data-led regulator, he explained this is at the core of its plans and it is doing work within this area.
Phoenix said: “There are two different lenses to look at. Data science - a comprehensive process that comes in many aspects of data processing according to regulatory compliance, visualisation.
“By contrast, AI is the implementation of a model to predict the future state of the world. It is increasingly used across financial markets. AI also helps in explaining how we use techniques for our own supervision, enforcement purposes, but also financial markets.”
He explained that advanced forms of data analytics, including AI can extract the insights that underpin regulators, the regulator’s decisions and enable the firm to be more efficient and targeted.
“Join me in being optimistic about the uses of data analytics, including AI [and how it will] improve classification and predictive accuracy of AI models, as well as the ability to automate sound decisions that benefit households and the economy.
“Consumers can access lower cost products and services and firms can realise efficiencies.”
However, the use of data and AI does not come without its challenges.
Phoenix explained that at the same time as it may lead to better situations, the speed and scale of AI systems can also amplify existing risks.
He said: “The complexity value means greater predictive accuracy and speed but also makes it harder to explain. Other challenges include a lack of transparency to explain the credibility of algorithmic decision making.
“This can make it difficult to understand how the AI outputs [came about] and can have unclear accountability and responsibility for decisions based on its outputs.”
He said algorithms cannot be blamed and it is important there is always somebody accountable for the decisions.
“We also need to have the force of digital inclusion in the community to approve these effects and perhaps most importantly, the challenges associated with potential bias may impact on different demographic groups.”
New data sources will come
During the discussion, panelists were asked on how they were using data within their organisations already and how far this will progress.
Rachel Lam, ombudsman leader at the Financial Ombudsman Service, said it is doing this in a number of ways in the way it regularly publishes information on the volume of complaints and uphold rates it is seeing.
“That is a really important source of information that we've been sharing for a long time now,” she said. "We know that drives behaviour, particularly products or services, especially around the uphold rates that we share.
“Equally, another source is the publishing of the decisions that we've issued. We know that is a really valuable source, we know that consumer groups use that as a source to understand how we speak about certain issues or complaints.”
She added: “We also publish insights that we see in the ombudsman news on our website and often that will come through us identifying a series of cases so when we work as a collective, we identify that we've seen a certain number of cases or a certain area where we are seeing high rates of accounts.
“When we see that, we do a deep dive to understand what issues we're seeing there and we'll share that after.”
Panelists were also asked how much additional data firms need to collect in order to beat the consumer duty expectations.
However, Phoenix argued that there was not “one magical piece of data” to solve all problems.
“There will be new data sources that come in the entire time. I'd say it is just going to be continued. We have to do transparency things all the time and each time a new piece of data comes in, we need to go back and retrain because it could change the mass that we've put into the models.
“So unfortunately, there is no single artist, this is only gonna get worse, it will get more complicated but that's exciting.”
sonia.rach@ft.com
What do you think about the issues raised by this story? Email us on FTAletters@ft.com to let us know
