If a computer turns down your insurance claim, ask for a person

On Monday a senior official at the Financial Conduct Authority set out, in public, the ways firms are getting artificial intelligence wrong.

Jessica Rusu, the regulator’s chief data, information and intelligence officer, said in her FCA speech that 86 per cent of financial firms now use AI. She described poor practice found in testing this year: tools built without a clearly defined need, vulnerable consumers excluded by default, capabilities overstated, and automated customer journeys deployed without enough testing. The good practice she pointed to included a clear path to speak to a person.

An insurance claim decision that you cannot question is not a decision you should accept.

That matters because claims are where the stakes are highest. Regulator figures reported by Which? in July suggest insurers reject almost three in ten home insurance claims, and the FCA itself flagged high complaint rates among home and travel claimants. Declined claims were the most common travel complaint reaching the ombudsman between April and June, according to its quarterly complaints data.

Here is the detail worth sitting with. A machine that reads claim forms faster can be a good thing. A machine that applies an old process faster, without anyone checking who it leaves behind, simply makes the old mistakes at scale. The regulator put it plainly: AI should improve consumer outcomes, not simply automate existing processes.

So if a claim is refused quickly, by text or app, with little explanation, ask three things. What exactly was the reason, in writing? Did a person review the decision? Has anything you told them about your circumstances, such as illness, bereavement or money worries, been taken into account?

Those are fair questions, and a well run insurer should be able to answer all three. If it cannot, that is your complaint, and the ombudsman is there if the answer still does not satisfy you.