Me again; I trust everyone is doing well and your loins are girded with breathable material, at least if you’re in the hot zone down South. I myself am considering wearing short sleeves as the mercury climbs to a heady 25 degrees, followed by a thing we get up here called ‘precipitation’. I may jet-wash some concrete as well, just because I can.
During the silly season / summer lull, I reckon one can tell a lot about what’s on our industry’s “mind” by the stories the trades run outside of any actual news happening. This is not due to any Druidic capabilities on my part; if you work in any kind of media title these days you know exactly what is generating clicks, dwell time, and comments. And while some may moan, the fact that every trade title is stuffed full of AI stories just now tells its own story; in much the same way as we get the government we deserve, we also get the media we engage with.
So while most big stories and themes wait for everyone to come back from the sunlounger, there is no shortage of AI content to think about. Most of it isn’t very interesting, but I thought you might enjoy this piece which has a bit more going on. It’s by a solicitor gadgie called Gareth from Birmingham, but despite that it’s absolutely worth a read.
Long story short, Gareth and his confreres asked a bunch of AI models for financial advice to see what would happen. Not financial planning; more the transactional end of things. They tested the results and considered whether they fall on the wrong side of the advice / guidance boundary. The piece is a bit of a tease, and we don’t quite know the results (wait for our white paper!), but the upshot is that there’s definitely some advice perimeter stuff to think about here for our friends in Stratford. Some AIs are being designed to provide proper, regulated financial advice and they’ll fall within the perimeter; more general ones aren’t and don’t and are much harder to regulate.
This may not be news in itself – many of you will have had a wee play with your favourite AI and may even have got right up there on your high horse when you find that you can get advice if you ask the right model the right questions. But it’s good to have someone testing the premise.
Personally I think it’s a given that the models deliver advice, even now. (Also a bit of a tease, but when we publish this year’s Advice Gap report in conjunction with Saturn, you’ll see some interesting stats.)
More interesting to me, then, is this: how consistent and how good is the advice? We all know that part of the regulated world exists to ensure that when client A presents with their unique situation and requirements, there should be at least some degree of consistency in terms of suitability even if they go to different advisers one after another. Adviser A might like a different platform, or a different MPS or a slightly different asset allocation to Adviser B, but it won’t be moonbeams vs potatoes.
It’s hard to measure this; you need quite a few elements to come together including the ability to run concurrent tests across different models, to have consistent client data but varying prompts and inputs, and also a base case of what good advice looks like. Ideally you’d do it with genuine clients to stop ‘advice creep’, but of course that brings its own challenges.
It’s the sort of thing that, were one to try it, would turn very quickly into a large-scale real-time AI financial advice lab experiment, and the only way you could do that is if you had quite a lot of advisers and industry professionals in a room, all at the same time, and a bunch of technology, and a history of doing really quite stupid things on stage, and probably a death wish…
…and this may be a decent time to mention that HomeGame 6 is at Leith Theatre in Edinburgh on October 7. Tickets available now. Will this event turn into a large-scale real-time AI financial advice lab experiment to check the quality and consistency of what these tools generate? How would people with a history of doing really quite stupid things on stage and a death wish go about this if it did?
I couldn’t possibly say. Yet. Watch this space.
Your music choice this week is dark and cold to make up for everything being bright and hot. You’ll be pleased to note Update favourites Insomnium are gearing up for a new album release, and here’s one of the first singles from it. Do we detect a bit of AI in the video? I think we do but the music is still doing what it’s mean to. Please enjoy Shadowlife.
