AI: the answer to everyday finances?

This article is the latest part of the FT’s Financial Literacy and Inclusion Campaign

Most ordinary consumers don’t have access to a personal financial adviser, but a growing number of AI services are promising to change that by offering tools to help spenders track their expenditure and give advice on how best to manage their money.

Although artificial intelligence is used in financial services and to manage more sophisticated financial products like ETFs, AI is increasingly being used for everyday “wallet management”, said Fabian Stephany, research lecturer in AI and work at the Oxford Internet Institute. This is typically where apps are granted access to the user’s bank account, analyse their spending patterns and offer them tailored guidance.

“You could say, this is my income and at the end of the month I’d like to save x amount of money, how can I achieve this?” he said. “But it could also compare you with your peers in a similar financial situation, and tell you if you’re spending much more, for example, on things like food or clothing.”

Some people are in need of a helping hand to steer their pocketbooks. According to the UK’s Financial Conduct Authority (FCA), 24 per cent of adults surveyed last year said they had low confidence in handling their money; 38 per cent said their knowledge of financial matters was low.

Apps such as Wally can track the user’s finances from across their current, savings, credit card and investment accounts to identify trends and predict upcoming expenses, as well as making chatbot-generated personalised spending plans when given an objective and a timeframe. Cleo, which plans to relaunch in the UK, can decide how much you could save on a monthly basis and automatically set it aside, as well as fine you a set amount — to go into a savings pot — if you spend too much on unnecessary or luxury purchases. Plum helps users to manage debts and overdrafts by setting aside an affordable amount as soon as they are paid.

Taking personalisation one step further, San Francisco-based start-up Wallet. AI, which plans to launch in the UK, will offer its users insights such as which days of the week they are mostly likely to overspend and — using Bluetooth and location data — whether they overspend in specific locations or with certain friends.

“We’re focused on behaviour and helping you create behavioural changes,” said chief executive Omar Green. “We all want to spend time with our friends, but if your budget is really tight you can maybe put off a pub date with that person who is a bad influence on your finances.”

Lenders such as Salad Money use AI to help borrowers who might otherwise fail to satisfy underwriting criteria. The social enterprise says that its use of AI democratises access to small loans for people with impaired credit scores.

Using individuals’ open banking data and machine learning to categorise their spending habits, it says that 85 per cent of applications are automatically accepted or declined, which means it can process more applications, and offer lower interest rates than its customers would typically have access to. Created as an alternative to payday loans, it offers a £10,000 loan repayable over 18 months at a representative annual percentage rate (APR) of 79.5 per cent.

It is unclear how they can monetise these apps — so people have to be really careful what they are signing up to

It has lent more than £47.5mn to 35,221 customers over the past four years. Just under half of its customers are NHS workers, while 71 per cent of its customers are female.

“It means that we can make better credit decisions on individuals who would typically be declined because they have had a missed payment, a county court judgment or have been made bankrupt in the last six years,” said chief executive Tim Rooney. “We push that to one side and ask ‘can this person take on a credit commitment?’ based on their ability to service it and their propensity to pay bills and debt.”

Some experts have concerns about the risk of AI in personal finance, such as whether the behavioural insights generated by transaction data and AI could be used to nudge people into spending more.

“There could be some bad uses of AI linking to personal finance and then being accessed by third parties to suggest to vulnerable consumers what they should do next,” said Rooney.

Many of the apps offer paid tiered subscription memberships, but also free services. Mick McAteer, co-director of The Financial Inclusion Centre, said he was worried that if apps have commercial relationships with lenders or insurance providers their targeted advice could be geared towards promoting their services.

“It is unclear how they can monetise these apps — so people have to be really careful what they are signing up to,” he said. “You could end up finding yourself limiting rather than expanding your choices.”

Although reputable companies may have data policies which require them to delete and anonymise data, users who are unclear how AI works may be less inclined to sign up for additional services reliant on the technology if they fear it will be used for nefarious purposes.

The FCA says that it is currently considering how regulation can promote safe and responsible use of AI in finance, and points to its Consumer Duty, which stipulates that companies must create products and services which provide good customer outcomes as a framework it will use to address issues arising from AI.

There is also a risk that financial services relying on AI could exclude people who aren’t digitally savvy. Banks increasingly use chatbots to steer help requests. While these can deal with increasingly complex questions, they often give limited and predefined answers, to the frustration of potentially distressed customers looking for help with their finances. 

Holly Mackay, chief executive of consumer website Boring Money, says AI could be helpful for dealing with people’s factual questions, such as whether they can have a cash and stocks and shares Isa, but for more substantive advice consumers may be unwilling to take financial advice from machines.

“People still overwhelmingly trust people and it will take a number of years for that to shift . . . I think this is a five-year evolution rather than an immediate revolution.”