The five AI principles that are worth sticking by.

2024 was a fascinating year for the development of AI and GenAI in particular. There was probably more debate about AI and GenAI than there was productive use of the technology. Just imagine what it will be like when we get AGenI (Artificial General Intelligence) and Quantum Computing.

And let’s not forget that 10 years ago, most Banks and FIs had banned AI as simply too dangerous; the release of ChatGPT4 to the man in the street slightly forced the hands of the Banks.

In particular, it was interesting to see how, tonally, the nature of the coverage changed in the marketing communications we all receive. In the first half of 2024, it was all about the ‘opportunity’; in the second half, it was all about the risks associated with AI. Indeed, there remain too many large Consultancies and Research Companies spending time creating FOMO, without actually delivering well-directed and timely input themselves. Understandably, they are looking to drive up demand for their own services, but in so doing they are contributing to the bubble effect. In reality, they are still struggling to formulate their best approach and that will only be derived on Industry experience i.e. what they can glean from their Clients.

We have prominent Industry figureheads who quite correctly suggest AI will be all about improving productivity for many people, at a ‘desktop’ level; but who also point out that the really big wins will be in Sales and Marketing, generally; Pharmaceutical R&D, solving major Engineering challenges and so forth – not really core Financial Services.

At MYRIAD we have a strongly held belief that adoption of AI will be driven by large Banks and other FIs determining their own AI path. This will involve the judicious selection of an AI partner and – in a completely sand-boxed environment – the exploration of just what is possible with carefully selected pools of data that have zero association with the outside world. Once proven, AI will then percolate through the organisation until such point it comes into contact with the outside world; just imagine. The Regulators will have a field day.

So, on to the Principles.

1. Get your Data strategy right; then get your Cyber strategy right and then – and only then – contemplate your AI strategy. Data leaks, data contamination and compromise will be the emerging stories in 2025.
2. The mantra of ‘crawl, walk, run’; it remains just as valid in 2025 as it was in 2024. There will be large Financial Institutions who find themselves trying to back out of AI projects that have been undertaken too quickly, that have accessed too much data and at too much cost, and these are the likely headlines in the New Year. Indeed, the advert that Salesforce ran in early 2024 – ‘If AI is the Wild West, is your data the new Gold?’ – was very well directed and remains so today, nearly a year later.
3. We stick with what we know; we will continue to be innovative and the clever technology will look after itself. The challenge we face as an organisation is sticking with our message: we have fantastic software which has been far ahead of the AI game for a very long time. We do not need to re-badge our technology as ‘AI’ or ‘GenAI’.
4. Use existing automation that is proven, safe and cost-effective. Automating the review and population of questionnaires has been available for years and does not require AI. Indeed, superior existing techniques for these activities remove the need to check and validate the work done by AI. It’s called software. One recent study begged the question: why halve the time, at six times the cost? The fact remains that the subject matter expert will never be 100% confident of AI output; after all, AI will always give you an answer.
5. Keep learning. We have adopted AI internally. It will not see the light of day until we are happy with it. As with all our coding (onshore), our migration to Cloud (careful) and our roll out to Clients (very careful indeed), so it will be with AI. We will test it and prove it, internally; we will roll it out internally, and only then will we open it up to Clients.

This is not a race to be first, it is a project to get right. That said, we will likely be a taker of AI, based on each Bank’s project, rather than a ‘maker’ of AI. We cannot conceive of an instance where external AI provided by a third party will be allowed unfettered access to a Client’s data. AI is data hungry; it will find a way.