Over the past couple of years, a number of small, digital banks have emerged on the UK’s financial scene, challenging the way things are done by the banking establishment.

Atom bank, headquartered in the North East, is one of these innovators.

It has global singer will.i.am on board as Strategic Board Adviser because of his interest in artificial intelligence and disruptive technology.

And last last year the bank teamed up with Newcastle University to ensure it is leading the way on consumer trust.

Atom is an app-only bank, allowing users to access products and services through their smartphone or tablet.

“Our intention is to be more secure – because app-based banking is more secure than internet banking – and to be more agile, efficient and therefore lower-cost,” says Edward Twiddy, Atom bank’s Chief Innovation Officer.

“There’s an asymmetry of trust, knowledge and power in banking. But when you’re new you can create all kinds of things differently. We want to have a set of values, a culture and a physical presence that is different.”

To help achieve this, Atom bank brought in the external expertise of Newcastle University to work on two projects – one looking at trust in financial services and one utilising blockchain to create secure and efficient banking products.

FinTrust, the first and largest of the projects, will help Atom learn what affects banking customers’ trust, form sound business ethics and create unbiased algorithms to power their products.

“Newcastle’s computer sciences department and expertise in big data and cyber security really do make them a world-class partner,” says Edward.

“Newcastle has depth in computer science in particular, but also a willingness to bring academics and researchers from other departments together.”

Combining skills

Academics from the Business School and the School of Computing are collaborating on the three-year £1.2m FinTrust project, which is funded by UK Research & Innovation.

They bring together researchers from various disciplines. Social sciences researchers are examining how trust can be measured, how trust is gained and lost, and the subsequent implications of this.

Researchers in Psychology and human factors consider trust on an individual level – in people’s experiences of the Atom bank app and reactions to financial distress.

Finally, the computing experts are studying bias within the algorithms that power AI and ways to make them fairer.

Professor Aad van Moorsel, Head of the Secure and Resilient Systems research group in the School of Computing, is one of three Newcastle University investigators on the team.

“The project is around understanding and improving trust of people who participate in financial services – using banks, bank cards, credit cards, mortgaging, lending and everything else,” he says.

“We want to understand how trust works in those situations, especially with financial services being automated and conducted online. What are the implications of that digitalisation and automation?”

The aftermath of the 2008 financial crisis and news stories about data breaches and biased AI systems mean banks have to work harder than ever.

Customers need reassurance that new products will work for them and that their data is being used correctly.

“I don’t think people have yet realised just how profound the rise of machine learning is and what that means for the relationship of trust between private individuals, corporations and money,” says Edward.

“Money is the most palpable form of trust, especially now that it can be a purely digital quantity.

“You’ve got to be mindful of the brand that you’re creating, the experience that you’re generating and how the customer perceives you, because these relationships are very easily broken.”

One step ahead

The two investigators leading FinTrust with Aad are Dr Karen Elliott, an expert in innovation and ethics at Newcastle University Business School, and Dr Kovila Coopamootoo, an expert in privacy and security from the University’s School of Computing.

Including people with a diversity of backgrounds as well as expertise is also vital to the project.

“It’s important to do these things across gender and background,” Aad explains. “Part of the bias in AI comes from the algorithms being designed mostly by men.

“Atom bank wants to be at the leading side of fighting against biases in AI algorithms. They want to be pre-emptive.”

Further down the line, the FinTrust findings will be turned into a Trust Engineering Toolkit – giving fintech service providers, regulators and consumer organisations a better way to measure trust and bias in automated financial advice.

But Atom bank is already using learnings from the research to inform new app developments. Being able to respond to the findings quickly and creatively is a great advantage of being a small, ambitious start-up.

Edward says: “One of the great things about being agile is that you don’t mind experimenting. You don’t mind seeing how learning and research can make you better.”