Artificial intelligence: crunching numbers

deagreez - stock.adobe.com

It’s that time of year when many intelligent people discuss the end of the tax year. But I’m unburdened by specialist knowledge on brilliant tax management strategies – or even that much intelligence – so let’s talk about something else.

We’ll stay on the subject of intelligence, but let’s make it artificial. You will have read about the march of our new robot overlords in many walks of life – from serious ones like AI healthcare, through to the fun game of getting Grok (X’s AI) to say bad things about Elon Musk.

It may not surprise you to learn that in your dealings with the financial services industry you may already be encountering AI, possibly without even knowing it.

I’m a bit sick of hearing about it in financial services; there is virtually no topic which some bright-eyed MBA can’t find an AI angle on. Yes, I am old and grumpy.

Behind the scenes, the places AI is cropping up in your daily financial lives aren’t earth-shattering, but are interesting. For example, an Edinburgh company has created an AI proposition which can scan phone calls to call centres for signs of client vulnerability and flag these for review. Imagine you’re someone whose job is quality assurance for a team of 200 phone operatives. You can’t listen to every minute of every call, so you start sampling, which will always be hit and miss, or need something to tell you what to look at – enter AI.

You may well get an advice report if you go to a financial planner. Many firms are now starting to use generative AI systems to sook (technical term) data in and produce the guts of the report. That normally takes an administrator up to a day to do, maybe more; AI can do it in no time, leaving the admin to check it over and look for “hallucinations” – we’ll come back to those.

Marketing emails from financial services providers may contain AI-generated articles. I don’t like these, they suffer from the “uncanny valley” problem where the AI product is good, but not the same as a person would do.

We also have interesting early-stage propositions which aren’t ready for the general public yet. Another Edinburgh firm is creating AI-driven investment models which pull in real-time economic data and trade automatically. Early days but potentially significant for how your money could be managed in future.

We have lots of clever stuff. But just because we have the ability to do these things, should we? And how do we make it a force for good?

I had the chance to be on a panel about ethical AI at Edinburgh University’s Business School last week, and it was an eye-opener. Models that “train” AI systems are based on stuff they are fed, usually from online sources. If you feed an AI model on car finance lending decisions, you’ll find that any biases or systemic flaws become codified or baked into the system.

But the flipside is true too. If you know that bad things are baked in, you can adjust algorithms to allow for them.

The currency for everything that AI can do is data, and that moves around at frightening speeds.

Whether it’s tax year end or not, or whether you’re bothered about AI or not, it’s good to always have a handle on where your sensitive personal data is and what’s happening to it.