Chris Ketcham reports: “State Farm, AIG and USAA have received preliminary approval from the Federal Aviation Administration to test drones for their claims and underwriting functions.” He muses on the wider implications of drone usage in insurance, focusing on what happens if or when drones see something they're not supposed to see. If a drone happens to “witness” a crime being committed, is the insurer obliged to report it? If a drone happens to see something in a neighborhood that could be interpreted as a symptom of a fraudulent claim, should the insurer act on that intelligence?
These are fascinating and important questions. You could widen the inquiry to include the image streams from proliferating fixed and mobile cameras, and the data streams being generated by the emerging Internet of Things. We're now in a position to collect great quantities of peripheral data.
But there's a major caveat. If your drone is programmed to survey, say, building damage, and it records details of an apparent crime taking place inside the building – who recognizes the apparent criminal activity? This is to assume a human agent reviews all the drone's video and takes decisions based on unstructured visual information.
While this may be the way the first generation or two of such systems work, we should expect additional automation. First, the building's smart devices (or the ones that aren't damaged) will communicate directly with the drone. Second, the drone will most like carry mapping tools that enable it to capture structural data as well as visual images. And whatever additional programming is built into the drone system, it will not contain the ability to recognize potential criminal activity unless there's a great advance in artificial intelligence.
Much of the data being generated today and in the future will be noise. It may be possible to use some of that noise to prospect for additional information. However, such nuggets will not appear spontaneously. People will have to look for them, and look hard. The issue, then, is not so much the possibility of an insurer finding out too much. It's whether or not the insurer cares to look hard. Or to sell its data by-products to organizations who know how to look hard. Insurance Thought Leadership
Comments