One of the things that has fascinated me most over the last few months has been watching how quickly AI recording tools such as Plaud have become part of everyday life.

When they first appeared, friends of mine laughed when I told them that I now record every meeting or conversation I have with the Plaud AI tool attached to the back of my phone. Surely you don’t want people seeing how you swear and joke around in meetings, do you? But they are completely missing the point, as my AI Twin already knows how I write up my notes, and not surprisingly, I leave the swear words out of a written summary.

But my pal Mark Sharples, who is a licensed FSA, was blown away when, after he talked me through his handwritten notes on my stepdaughter’s mortgage, I was able to summarise them and send him back a complete set of meeting notes before he even reached for his next beer on the sun lounger next to me in Mauritius.

Like most people, he saw these devices simply as clever note-taking tools that could record meetings, summarise conversations and save people from scribbling pages of notes, but he did not see them as something for him. However, this one meeting summary changed his mind.

However, his next thought was “Christ, how am I going to get compliance to sign this off?” and do I even want to, as this may be the biggest weapon ever handed to a compliance officer with a grudge, because no human is ever 100% compliant with every word or phrase we say.

But the bottom line is that it is better for both the IFA and the customer because every conversation is documented and set out in clear English, which can prevent many misunderstandings.

Also, not having to constantly take notes improves the human interaction of the meeting, with the IFA being able to listen to everything the customer says or often does not say, allowing them to get a better feel of what the customer wants to achieve.

There is another benefit, which I think is even more significant, as most consumers in my experience do not ask enough questions because they simply do not understand the language being used and don’t want to feel shown up.

Financial services are full of terminology that professionals use every day but which can sound intimidating to everyone else. Historically, many people have nodded politely, signed the paperwork and hoped they had understood enough to make the right decision.

However, they can now leave the meeting, ask AI to explain every technical term in plain English, compare recommendations with alternative products, identify areas they forgot to ask about, and return far better informed than would ever have been possible in the past. Knowledge, once again, becomes available to everyone rather than only to the expert sitting across the desk.

But as Mark worried, I also worry that many institutions may end up using AI in the wrong way because the same technology that empowers consumers can tempt them to become increasingly defensive.

If every conversation is analysed, every sentence scored for compliance, every adviser judged against thousands of regulatory rules, and every interaction monitored by AI for potential breaches, there is a danger that people stop having genuine conversations and instead begin reading scripts designed primarily to satisfy regulators.

The customer may technically receive every disclosure required by law whilst leaving the meeting with less understanding than ever, because the adviser has become more concerned with saying the correct words than with ensuring the customer genuinely understands them.

I have seen similar things happen throughout my career.

This was particularly the case at Rock Insurance, when we implemented voice recording of every call. The aim was to capture the best skills of the best salespeople and to show all staff how to close, but what actually happened was that our compliance officer became a little Hitler with newfound power and produced a set of script-reading dummies who were 100% compliant but converted at 32% less because customers were simply bored by the process.

AI gives us an opportunity to avoid making the same mistake again, but I have a nasty feeling many businesses will.

Instead of using artificial intelligence solely to identify mistakes, imagine using it to detect confusion. Rather than asking whether every regulatory statement was read aloud, ask whether the customer genuinely understood the explanation.

Rather than measuring how closely an adviser followed a script, measure whether the customer left the meeting feeling informed, confident and able to explain the recommendation in their own words. These are very different objectives, but only one genuinely empowers consumers.

Technology has always reflected the intentions of those deploying it, and the internet could be used to educate or mislead, while social media could bring people together or drive them further apart.

AI will be no different. Used well, it can create the most transparent relationship ever between financial institutions and their customers, where both sides have access to the same information and leave conversations with greater confidence. Used badly, it simply becomes another surveillance tool that encourages organisations to optimise for regulatory compliance rather than for customer understanding.

The real opportunity is not to use AI to prove that institutions have said the right things, but to ensure that customers have heard, understood and benefited from those conversations. If we get that balance right, AI could become one of the greatest consumer protection tools ever created.

However, if we get it wrong, we will simply have replaced human judgement with digital bureaucracy, and that would be a remarkable waste of one of the most empowering technologies any of us are likely to see in our lifetime.