AI may be transforming financial advice, but the rush to adopt it carries risks as well as opportunities. Nick Ryan, CEO of Pilot, argues that advisers should look beyond the hype and focus on using AI where it genuinely adds value without compromising judgement, accuracy or the craft of advice.
At Pilot, we’re building technology for advisers, so almost by definition we are not anti-AI.
However, what we also very much are is anti-hype, and we think the message driving AI adoption right now – essentially: don’t get left behind – is ill thought through at best, for a couple of important reasons.
Fish and bicycles
By way of scene-setting, in 1990, Steve Jobs famously remarked that a computer is “the most remarkable tool that we’ve ever come up with, and it’s the equivalent of a bicycle for our minds.”
Now here we are in 2026, rushing to embrace our latest mind-bicycle, namely AI, or get left behind with the computational equivalent of a Raleigh Grifter.
However, way back in 1970, Irina Dunn equally famously remarked that “a woman needs a man like a fish needs a bicycle.” In other words, fish simply don’t need bicycles, whether they’re the latest carbon-fibre superbike or not.
This tension is at the heart of the AI conversation.
As the co-founder of a software company, I’m no Luddite, far from it. But given that I’m also an adviser, my question to you is not just whether AI is a bicycle for your mind; it’s also whether you’re acting like a misguided fish.
Mundane in, unhinged out
Point number one: we’ve all heard the ‘garbage in, garbage out’ paradigm.
Well, think of a guy. He’s a practising IFA and also the co-founder of a financial planning software company. It helps if you think of him as incredibly handsome, but it’s not essential.
Recently, this guy (we’ll call him Rick) asked a popular AI (we’ll call it RunDMT) to help format some numbers into a graph in Excel. What he got was a complete stocktake for a real shop in Chicago. Rick is here in the UK, and hadn’t even mentioned Chicago, nor asked about stocktakes.
But the AI gave him one anyway, delivered with its customary air of obsequious confidence. Of course, Rick could easily see right away that his friend RunDMT was off the reservation here; this was an obvious and extreme case of mundane in, unhinged out.
Confidently wrong
Nevertheless, confident wrongness is a significant problem. There’s a good reason we’ve traditionally abhorred it when we’ve encountered it in the kind of overpromoted dolts we’ve all had to work with at one time or another.
AI, similarly, doesn’t fail the way that good humans fail; that is, hesitantly, displaying visible uncertainty, maybe even conspicuously perspiring. When AI fails, it fails in complete, eloquent sentences, with proper formatting.
Do you want an overpromoted dolt that’s subtly-but-consequentially wrong working in your practice? How confident are you that you’re picking up every error? The FCA does not accept “the algorithm hallucinated” as a defence, and neither will your PI insurer for that matter…
Hoomans
You’re probably thinking this piece is about to go the way of the twelvety billion other articles on AI swilling around the trade discourse right now. You know the ones: “AI is just a tool, it’s us hoomans who use our judgment and know when our clients are sad, etc.”
Well, I don’t really disagree. As a software company, though, we’re trying to cleave to certain principles governing its deployment as a tool: good data in, single source of truth maintained within our environment, well-defined tasks, and so on.
Because urgency isn’t a strategy. Firms rushing to bolt AI tools onto everything might well find they’ve simply added automated error and inaccuracy. And whilst it remains ‘just a tool’ for the time being, it’s still on us to decide if it’s the right tool for the right job.
More on fish and bicycles
However, notwithstanding our natural empathic capacities and all the rest of it, somebody more paranoid than me might already be worrying just how long it’ll be until the bicycle realises the fish sitting in the saddle serves no useful purpose.
As advisers, we are genuinely worried that one day it might well be as good as us at the ‘hooman’ bits. And that raises what I think is the second important, slightly submerged issue.
Financial planning is a craft, at least of sorts, a characteristic it shares with virtually every other human endeavour. When we spend our time – by which I mean the hours and days of our lives – wrestling with a client’s situation, turning the problems over, thinking them through, we are doing something that matters beyond the output: we are doing our work.
There’s a plausible version not just of advice but of many professions’ futures where we spend our days simply prompting AI, reviewing AI output, correcting AI errors, and prompting AI to try again. It feels like that’s already begun.
I ask you: where’s the satisfaction there? We’d all like to be Steve Jobs. What I don’t want to be is a fish effectively minding the bicycle. Let’s be careful what we wish for.
