Can AI close the advice gap?

At the moment, there is a wide ‘advice gap’: many people would like professional financial advice but either can’t afford it or feel it might not be good value. According to a report from financial consultancy The Lang Cat entitled ‘The Advice Gap 2024’, only 9% of consumers paid for professional financial advice in the two years up to July 2024, down 2 percentage points from the 2023 findings. Of those consumers who did pay for financial advice, 91% found it helpful.

In the meantime, many people are busy looking for free financial tips online, but that doesn’t necessarily end well. As the Financial Conduct Authority (FCA) recently pointed out, some so-called ‘finfluencers’ illegally promote unauthorised schemes that are totally unsuitable for retail investors. There is also no shortage of outright scams. According to Citizens Advice, one in five people fell victim to financial scams in 2023 – three-quarters of which were found on social media sites.

But the number of bad actors online doesn’t mean that the internet can’t be a useful source of financial information. Citizens Advice, for example, has helpful tips on tackling financial problems such as unmanageable debt. People struggling with debt can also speak to its regulated debt advisers.

However, resources such as Citizens Advice require people who may be feeling stressed or worried to read online information carefully, or to talk to someone about problems they may prefer to keep private. Artificial intelligence (AI) bots such as ChatGPT promise an easier way forward. AI will answer almost any question promptly, will not judge, appears completely private and may even get things right.

Could AI then be the means of closing the advice gap and potentially replace the provision of regulated financial advice by humans?

“Technical financial advice will be free 10 years from now and AI is the greatest opportunity for, and the greatest threat to, financial advice,” says Ian McKenna, Chief Executive of AdviserSoftware.com.

Although technology is going to change the provision of financial advice, just as it is changing almost every other aspect of the economy, not everyone agrees that the transformation will be so radical.

“Consumers want fast, efficient service at their fingertips at any time of the day,” says Prashant Jajodia, Managing Partner, Financial Services Sector Leader, at IBM Consulting. “What GenAI can do is trawl through lots of info. So can it provide low-level advice, yes, but there is also the regulatory aspect.”

Jajodia says that IBM has three rules on the implementation of AI: it is there to augment humans, not replace them; the data insights belong to their creator; and the output must be transparent and explainable. Given that, “AI cannot replace humans in providing financial advice,” he says. “What AI can do is augment and assist human advisers with good insights.”

That will be what the regulator wants to see, particularly following the introduction of Consumer Duty rules in July 2023. “Consumer Duty really is a step up from previous attempts to deliver good outcomes,” says Guy Sagar, Managing Director at Promontory Financial Group, an IBM subsidiary. “It can be encapsulated in ‘the avoidance of foreseeable harm’, which changes the game. The regulators are also very clear on firms creating good value for consumers.”

But providing good value for consumers generally involves driving down costs at service providers. McKenna says that an adviser now can support between 80 and 120 customers at most, but there is technology coming that will enable one adviser to support up to 1,000 customers while remaining compliant with regulatory requirements.

That could blindside some. A recent survey of 200 independent financial advisers by Opinium, a market research agency, found that 23% intend to start using AI tools in client services in the next year, with around 14% already using them. However, just over half expect AI not to impact on their firm and around 45% think it will have neither a positive nor a negative impact on the industry.

McKenna argues that “10 years from now, financial advice will be far more accessible, but people will not necessarily realise they are getting financial advice because it is embedded in a lifestyle management system”, also known as a digital financial assistant.

Jajodia, although he also sees scale as important, does not believe that the end result in the UK or Europe will necessarily be one super app. “When you look at super apps, they haven’t caught on as much here as in emerging markets,” he says. “I think a super app will be based on geographical and cultural habits… India, for example, is more open to disrupting the market.”

That may be partly because India is a young society. The median age is just 28, compared with 40 in the UK. Will it be possible in an ageing society to expect elderly consumers to rely on digital interfaces? “We’ll need to have analogue services to support people in that situation,” says McKenna. “But I know of one adviser who has gone through a six-month exercise to get all their clients using their online portal. It can be done.”

McKenna argues that, in the future, the role of the adviser will be to think and understand ‘which of my clients is most likely to react badly to a given set of circumstances?’.

“We are seeing new technology emerging that will help advisers to understand the psychology of their clients far better,” he says. “For example, if somebody is a visual learner and you send them a 40-page text document, they ain’t reading it. The technology can match a visual learner with an adviser who is also a visual learner and it can also take text-based information and create visual content.”

Existing use cases for AI in financial advice include summarising the information that an adviser has taken from a client meeting. “Reading, classification and the drafting of a response are things that AI can do rapidly,” says Jajodia.

But much as people enjoy speed and convenience, they also want to feel safe. Jajodia points out that privacy needs careful consideration in GenAI. “Good design and implementation practices are all about how we ensure that privacy is maintained,” he says. “Sensitive data should only be available to those authorised to have it.”

Another fundamental aspect of feeling safe for most of us, says John Somerville, Director of Financial Services at LIBF, is having reassurance from another person that we’re doing the right thing. “If that weren’t the case, everyone would have a will, a power of attorney and some life cover as all these things are easy to do and relatively inexpensive,” he says.

Importantly, even if much can be done without an obvious human touch, the models that GenAI uses to provide financial advice will need access to the right data. “Large language models probably already have around 95% of the data on the internet in them but less than 1% of enterprise data,” says Jajodia. “Our mission is to bring in enterprise data. If models are trained on bank data, they are much more efficient at providing advice.”

Jajodia doesn’t see open data – in which consumers will be able to share their data from a wide range of firms, including bigtech companies – as a game-changer in building good consumer finance models. “Open data would make LLMs [large language models] more powerful, but it’s not a showstopper,” he says. “Firms would also be ingesting their own data.”

Will human financial advisers still be needed in 10 years’ time? “They will be,” says McKenna. “There will be lots of people involved in the creation of financial advice. Few of them will be directly customer-facing, but there’ll be a lot working behind the scenes. Ultimately, the machines will be able to do far more than we can.”