Ever wondered if you can afford to take early retirement? Traditionally, you might have engaged a financial adviser to help explore your options. This week, I’ve been talking to one reader who’s been using ChatGPT instead.
Unlike the 12mn Britons estimated to be in the “advice gap”, 58-year-old Jim can afford to pay for regulated financial advice. With a £2mn defined contribution pension entirely invested in global equities, he has a further £850,000 in liquid assets (primarily fixed income) plus a £2.5mn home. He turned to generative AI out of curiosity, but the results surprised him so much, he shared them with me. So how worried should advisers be?
Jim provided ChatGPT with a synopsis of his financial position stating his desire to retire at 60, asking: “Do I need to change the profile of my DC pension fund when I retire?” Within seconds, it generated an easy-to-read three page summary identifying that Jim’s central dilemma was managing a sequence of returns risk — the danger that a market downturn early in his retirement will reduce his portfolio’s longevity, especially if he’s drawing an income from age 60.
It listed the pros and cons of keeping his pension fully invested in equities; moving to a more balanced 60:40 portfolio of equities and bonds or adopting a “bucket strategy” with three years’ income in cash, 5-10 years in moderately risky investments and the rest in equities for long-term growth.
Next, it urged Jim to explore his cash flow needs, noting his large fixed income pot meant he could afford to keep his pension more aggressively invested, and the longevity risk of a 30-plus year retirement was flagged.
Not bad — but the next heading of “recommendation” made me gasp. Would anyone rely solely on the word of an unregulated chatbot? This concluded Jim should gradually de-risk and reassess his position annually, though the “final tips” section (with a yellow warning icon) did say “consult a financial adviser before making any changes”. And just as well, as ChatGPT’s confident assertion that DC pensions are outside your estate for inheritance tax purposes will not be true when the law changes in 2027.
The AI then prompted Jim to see if he’d like to see a projection of his income in retirement under different drawdown and asset allocation scenarios, and a Monte Carlo simulation duly appeared. Wow. How about playing around with healthcare cost assumptions? Tempting — but he had now hit the free usage limit and would have to pay $20 per month to continue. “Maybe this is the new cost of financial advice,” he says.
The information Jim surfaced in 15 minutes was nothing that he couldn’t have found online with a few hours of dedicated Googling. But AI’s ability to summarise and distil this speedily into a personalised jargon-free report, with added interactive prompts, was impressive.
I’m relieved that Jim now intends to seek advice from a fully regulated human being, but advisers think his ChatGPT experience will have been a great primer for this.
David Hearne, a chartered financial planner at FPP, thinks that someone who starts with a question in ChatGPT is more likely to end up speaking to an adviser than someone who asks no questions at all.
“The more informed someone is when they come to us for advice, the better,” he says. However, Hearne and other advisers were quick to point out AI’s limitations.
When drawing up a financial plan, they would ask broader questions about Jim’s health, his family circumstances and estate planning goals, plus his likely income needs in retirement and how to sequence these in the most tax-efficient way.
“What the AI doesn’t say — yet — is ‘I can’t answer this until you tell me that’,” says Hearne. Much rests on an individual’s prompting skills, though the ability to ask: “What haven’t I asked that other people in similar circumstances have done?” is a tantalising possibility.
“Despite AI’s obvious risks, it’s still a heck of a lot better than some TikToker saying you can retire at 40 if you pay for his trading strategy,” says Adam Walkom, co-founder of Permanent Wealth Partners.
Like many others, his financial planning firm already uses AI tools to streamline admin, transcribe and generate action points from client meetings, and help visualise analysis of client portfolios. Advice firms are most definitely alive to developing their own AI-powered advice generating models in future, but cautiously await more guidance from the financial regulator.
“The typical adviser oversees 100 clients,” Walkom says. “If someone started a firm providing AI-enabled advice that was sanity checked by a qualified adviser, maybe they could oversee 1,000 clients.” With regulatory blessing, this could open up a much lower cost advice model, though he stresses protecting and anonymising client’s financial data would be “the absolute starting point”.
For its part, the Financial Conduct Authority is keen to explore AI’s potential. It launched a tie-up with Nvidia this week, creating a “supercharged sandbox” for financial services firms to experiment with AI in a controlled environment. Just one of the ways it is helping firms navigate AI adoption, the FCA intends to apply its existing regulatory framework to AI rather than create an additional one.
But just as it is clamping down on finfluencers, does the regulator need to better protect consumers who are already turning to generative AI?
As an experiment, I asked a more mass market question — what should a 50-year old woman earning £30,000 with a £80,000 DC pension do to better prepare for retirement? I was served a fantastically useful summary referencing the PLSA’s retirement living standards to estimate what income I’d need; told to check my state pension forecast; shown examples of how much more I’d need on top to close the gap and given a link to the government’s MoneyHelper pension calculator to work this out.
Although I wasn’t told that I could sign up for a free Pension Wise appointment, which provides guidance from a qualified professional, doing all of the above would have made me much better prepared for one.
Generative AI clearly has the potential to transform financial education and deepen consumer engagement, but it is not infallible. Although I think the pros massively outweigh the cons, the backstop of qualified human guidance is very much needed. Seeing the ease with which my 20-something stepchildren interact with AI chatbots on their phones, their generation may end up better prepared for retirement than mine. But that’s assuming AI doesn’t put us all out of a job.
Claer Barrett is the FT’s consumer editor and author of the FT’s Sort Your Financial Life Out newsletter series; [email protected]; Instagram and TikTok @ClaerB [email protected]