Banking is boring. AI doesn’t mind.

Most people know they should compare savings accounts, shop around for better insurance or review their pension. They rarely do. Why?

David Birch
Author, advisor and international speaker

Because it’s boring. A lot of business models are based on consumer inertia and cognitive fatigue. Most people simply can't be bothered to research the different options and get the best deal possible. Take something as simple as renewing your car insurance. Most people get the renewal notice and think, "I really should compare it with quotes from a few other companies." Then they remember they'll have to answer the same questions twenty times, compare all the quotes, and work out which one is actually better. They hit "Renew"—not because it's the best deal, but because they don't have to think about it anymore.

Banking is full of examples like that. People leave money sitting in accounts earning little or no interest. They keep using the same payment cards. They don't review their pensions. They don't switch savings products. None of these decisions is especially hard. They're just tedious.

But if I have an AI working for me, it doesn't get bored or tired. It will happily compare savings accounts, review insurance policies, look for a lower mortgage rate while people go about their daily business. It’s convenient. And convenience always wins.

People sometimes tell me, "I'd never let a bot move my money around." They say that today, because such bots don’t exist yet. But when the bot quietly appears one day, and says, "I've found a better savings account and moved your money. You're earning another £500 a year," people will quickly decide that's a service they like.

The point isn't that AI is making brilliant financial decisions. But it's removing the friction that prevents people from making better ones.

Banks expect AI to make them more efficient. Is that really the big change that’s coming?

No. The big change isn’t that banks are getting AI. It's that their customers are getting it.

When I help banks think about AI strategy, the conversations often start in the same place: How do we use it to reduce costs? How do we automate customer service and improve marketing?

Those are perfectly reasonable questions, but they're incremental. The real question is: what happens when every customer has an AI assistant acting on their behalf? That's a different world.

Banks have spent decades designing products for humans. Now, they have to start thinking about customers who delegate routine decisions to software.

So this isn't really about AI replacing people. It's about people delegating work they don't enjoy doing.

Exactly. We already delegate all sorts of things. We hire accountants to do our taxes. We let GPS choose the quickest route to our destination. We delegate these tasks to others, not because we can’t do them, but because we don’t want to.

Financial decisions are simply another category of work that people would rather not do.

Banks sometimes act like their customers wake up in the morning excited about financial products. But do you really think somebody who already has a virtual life coach, a virtual fitness coach, and maybe even a virtual boyfriend is going to want human contact from banks just because they love banking? Of course not. If they have an AI they trust, they'll let it handle routine financial decisions.

What does that mean for the industry?

It means a lot of today's business models become harder to sustain. Take deposits, for example. In the UK, there's an enormous amount of money just sitting in accounts, earning no interest at all. That's great for banks. They can use depositors’ money without having to pay interest.

But an AI won't leave money sitting there, doing nothing. It doesn't care whether the better rate is half a percent higher or one percent higher. It doesn't think, "I'll get around to that next weekend." It just moves the money.

Now imagine millions of customers doing that – automatically. Suddenly, a significant source of the banks' profitability comes under pressure. That's why I describe this change as “existential.”

If AI starts making financial decisions on our behalf, what happens to traditional banking brands?

I think brands will still matter, but perhaps not in the way banks imagine.

Today, banks compete for people's attention. Tomorrow, they may increasingly be competing for the attention of their customers' AI. That's a very different problem.

If my AI is choosing financial products on my behalf, it isn't looking at television commercials or glossy marketing campaigns. It's evaluating price, trust, performance, and whatever preferences I've given it. In that world, the customer relationship starts to look very different.

So, instead of building relationships directly with people, they may find themselves increasingly providing services through an AI that’s acting on the customer's behalf.

How quickly could this happen?

Probably faster than most banks expect. Technology doesn't have to replace every customer overnight. It only has to become good enough that people start delegating more and more routine decisions. Once that process starts, it feeds on itself.

People discover that the AI really did find a better insurance policy. It really did increase the return on their savings. It really did spot a better mortgage rate. Trust grows; consumers delegate more tasks. And suddenly, banks discover they're no longer dealing with customers who occasionally compare products. They're dealing with customers whose AI compares products continuously.

Nobody really has a strategy for dealing with customers who suddenly become 1,000 times smarter, literally overnight.

If AI starts making financial decisions for us, does that not increase the risk of fraud – instead of cheating people, scammers start targeting their agents?

That's what most people assume. My instinct is that the opposite may prove true.

The reason people get scammed is that they can be fooled. I can pretend to be Brad Pitt. I can create convincing video and photos; I can produce messages that appear to come from his social media account. Last year, a woman in France sent about $850,000 to someone she genuinely believed was Brad Pitt.

Humans are vulnerable because we judge appearances and respond to things emotionally. Machines don't work that way. An AI doesn't ask whether something looks convincing. It asks whether the credentials are valid. I can't forge Brad Pitt's digital signature. A cryptographic signature either verifies or it doesn't.

Sure, criminals will use AI, too. Fraud will evolve, as it always does. But if we're talking about social engineering—the kind of fraud that persuades people to do something they shouldn't—AI could actually make us much harder to fool.

So the challenge isn't really AI. It's identity.

Exactly. As AI becomes more capable, identity gets more important. Today, when I log on to my bank’s web site , the bank wants to know that it’s really me. Tomorrow, it may need to know that the AI acting on my behalf is really my AI. So, the infrastructure we really need isn't simply more AI, it's trustworthy digital identity.

We need ways for people, organizations and, eventually, software agents to prove who they are and what they're authorized to do. Without that, none of this works.

What does that architecture of trust actually look like?

People sometimes imagine it as one enormous global identity database. But I don't think that's where we're heading. A better comparison is the mobile phone network.

I can land almost anywhere in the world, switch on my phone, and it works. Why? Because the underlying standards are shared. The security is built into the system. The different operators don't all have identical databases, but they agree on common protocols.

Identity could evolve in much the same way. Different organizations could issue trusted digital credentials that work across different countries and different services because everyone agrees on the standards. That's a much more realistic vision than trying to build one universal identity system.

Some people hear that and immediately worry about surveillance and privacy.

They should. Privacy matters enormously. The encouraging thing is that modern cryptography gives us ways to prove things without revealing all the underlying information.

Imagine I'm sending money to you. Today, we often transmit my personal information and then ask the receiving institution to check whether I'm on a sanctions list, or whether I'm allowed to make the payment.

But maybe I don't need to send you all my personal data. Instead, a trusted organization could verify my details and give me a cryptographic proof showing that my background checks out.

What gets transmitted isn't my personal information, but the proof that my information has already been verified. That's a much more privacy-enhancing way of thinking about digital identity.

It sounds as though you're describing a future where trust moves away from people and towards credentials.

That's a good way of putting it. For hundreds of years, we've relied on human judgment: Does this person look genuine? Does this document seem authentic? Does this signature appear to match?

Increasingly, computers can answer those questions much more reliably than people can.

That doesn't eliminate trust. It changes what we’re trusting. Instead of appearances, we trust cryptography. Instead of believing what someone claims, we check credentials that have been vetted by trusted organizations. That's a much more robust foundation for a digital economy.

If you look five or 10 years ahead, what's the biggest mistake banks are making today?

I think too many banks still see AI primarily as a way of becoming more efficient.

Of course, they'll automate operations, reduce costs, and improve customer service.

But the real transformation starts when customers begin delegating routine financial decisions to AI they trust. Once that happens, banks are no longer just competing with other banks. They're competing to become the service provider chosen by somebody else's AI.

That's the shift I think many people still underestimate. The future of banking won't be decided only by which banks have the best AI. It will also be decided by which banks AI chooses on behalf of customers.