I, FA: Artificial Intelligence in the advice process

Writing articles about Artificial Intelligence in advice is like trying to hit a target moving at Mach 3.

I last went here when ChatGPT was a twinkle in some vitamin D deprived Frankenstein-dev’s eye. There's a serious risk that the next leap in AI capability will make me look stupid before anyone’s even read this. But I’m fully committed to public acts of stupidity so I’m going for it.

See, the thing is, I still don’t think much has changed. Moravec’s Paradox still applies: “It is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility.”

Every human possesses complex sensorimotor and social skills developed over millions of years, which they use without thinking about it. Even people who vote for Donald Trump and subscribe to Russell Brand’s YouTube channel.

These less-obviously-impressive social tasks are harder for AI to replicate than apparently complex ‘thinking’ tasks. Hence, we still don’t have robot waiters, but Deep Blue spanked Garry Kasparov at chess using dumb brute force. Programmers had initially thought the key to machine chess was the ‘Type B’ approach (which tries to think like a human, using heuristics to narrow the options down to the more plausible moves). Turned out, it was better to just sit back as a computer the size of your living room considers all 1045 possible moves in its monstrous, soulless way. One nil to the Robot Overlords. Jurassic Park!

But that was a millennium ago in AI years, and things have moved on

ChatGPT is vastly more sophisticated. It’s smashed the Turing test. It’s got some good parlour tricks, such as interpreting illiterate gobbledygook - it’s got ‘typoglycemia’ which does feel like a creepily human quirk.

What does this all say about the imminent death-by-T1000 of professional services, and financial planning in particular? Having spent as much time working in law firms and accountants as financial planning practices, I’m confident that we have less to worry about than them. What feels like an inconvenient truth in the context of Consumer Duty is a strength in the context of the threat of AI: our value is hard to quantify.

Sure, sometimes it isn’t. We’ve all had cases where we save someone a gazillion handily identifiable pounds of IHT with our clever planning, then runoff to tell everyone on X like the needy little things we are. But more often we add value in ways that are profound, but annoyingly intangible.

Investing and background administration will become more automated and commoditised, but that’s good, and it was happening anyway. Soft skills will still be where the real value is. We’re waiters as much as chess players.

These soft skills are much more important to us than other professions because we need to establish a good, close relation to do the job properly. The key currencies in financial planning are relationship building, gaining trust, influencing. These are often dismissed as sales skills, but there’s much more going on. Advisers (both the old guard and new model variant) are mostly really good at these things.

Ongoing engagements mean we loom sinisterly in the background, stopping people doing the tempting-but-dumb things investors love doing. Like selling everything at the worst possible time when their instincts are screaming at them to do it. Most clients understand the logic of buy low / sell high, but seeing it written down apparently doesn’t stop them getting it catastrophically wrong all the time. Why would having it regurgitated by an AI programme be any different? ‘Peace of mind’ has been done to death as a concept, but it happens to be a real thing, which we have the power to deliver more than other professional services.

Financial planning is also largely about slippery trade-offs. There’s no such thing as an optimally tax-efficient strategy. There’s a choice between maximising tax savings in your lifetime or for the beneficiaries of your estate. Retirement planning is about finding a suitable trade-off between spending money now and later in your timeline. Most people want some vague middle option that is subjective to them, and probably changes over time. They greatly value help articulating and prioritising their goals. Planners need to be navigators at the crossroads where the science of investing meets the art of client goal setting. As Pablo Picasso put it, “What good are computers? They can only give you answers”.

‍

This article appeared in Nucleus: Illuminate on 26th September 2023:

https://illuminate.nucleusfinancial.com/blog/financial-planning-will-always-be-beyond-ais-reach

‍