Connect with us

Artificial Intelligence

When common sense beats AI

A short-term insurance company finds technology and AI more effective for internal than for external use, writes ARTHUR GOLDSTUCK

Much is made of the use of artificial intelligence in the short-term insurance industry to detect fraudulent claims. In this context, vendors of AI software like to cite its use in the analysis of voice patterns and facial cues in the kind of video submissions that have become commonplace in making insurance claims.

However, the reality is somewhat more mundane. King Price, a short-term insurance company that celebrated its 10th birthday this month, says that it finds technology and AI more effective for internal than for external use.

“We have tools to measure productivity,” says Gideon Galloway, founder and CEO. “I can see when a programmer codes or just sits all day long, although our culture was never clock-watching. It was always output-driven and about productivity. On the efficiency side, it looks at QA (Quality Assurance ). Our IT systems transcribe everything and it starts picking up if, let’s say, a consultant or being aggressive to a client.”

Naturally, fraud prevention does enter the technology picture, but not in the way the media hype would suggest.

“We use AI as one of our fraud tools internally,” says Galloway. “I’m not talking about which claims we reject or repudiate. This tool looks for anomalies. For example, if one consultant keeps passing on to the same consultant on the other side, it says here’s something funny, just look at it.

“Then you can look at it and maybe it’s normal, that’s the normal chain of command, and that’s fine. But maybe it’s two people colluding. It also picks up anomalies in terms of syndicates, like collusion by panelbeating groups.”

The reality is that video assessments are old hat – and not as reliable or even as useful as the hype suggests.

“We could do photo or video assessments 10 years ago. The submission is still plugged into your back-end IT system and you make decisions based on that. Most of the front-end stuff that you see today is literally just a video that comes to the operator inside the insurer, but the claim is not being decided on the video. A human may be looking at this video and in the end, probably does not even base a decision on the video. They probably phone up the panelbeater to find out more and then do a manual assessment at the end.

“In our early days there was a panelbeating syndicate that would make chalk marks on the car, and on camera, it looks like this car has been in a serious accident and we need to write it off. And then you pick up that it was just chalk and there’s nothing wrong with the car, for example. So there are still flaws in some of these technologies, although obviously, technology does speed up the customer journey. But it goes far wider than just videos and photos.”

AI does help to settle claims faster, but that typically happens at the moment the claim is submitted.

“We are already approving claims within a second so it doesn’t even have to take the video into account. I mean, if you’re a paying customer and over the years have proven certain things as a customer, the AI tool will decide immediately, and it doesn’t really matter what you submitted. We would pay it in a second. It could be your TV, car or anything like that.”

The real secret, says King Price deputy CEO Rhett Finch, is not AI, but common sense.

“We settle up to 20% of our claims now within a handful of seconds. A perfect example is a glass claim that comes through and can be assessed quickly, because it’s a paying customer who has been around for a long time, and then it’s automatically approved. An internal goal is to get that close to 40% for certain claims. It makes a huge difference to the user experience.”

* Goldstuck is founder of World Wide Worx and editor-in-chief of Gadget.co.za. Follow him on Twitter on @art2gee

Subscribe to our free newsletter
To Top