Could AI Transform Insurance Ethics?

Could AI be used by regulators to test how committed insurance executives are to building trust with policyholders? Artificial intelligence is transforming the relationship between insurer and insured. And it’s now being used in ways that could transform the relationship between insurer and regulator. It has implications for public trust and executive careers. It has emerged that a large investment management firm has been using an AI-based form of voice analysis to test the confidence of the chief executives of the firms in which the firm has significant holdings. Called "affect analysis," it’s being used to detect any disconnects between what the chief executive is saying and the level of confidence with which she's saying it. The feedback could be used to pinpoint weaknesses around which further questions are raised, or to just automatically adjust an investment or research recommendation. See also: Strategist’s Guide to Artificial Intelligence The idea behind this approach should not be something new to insurers. They’ve been using it for some time to analyze how claimants describe the circumstances of their loss, looking for indicators in their voice of a potential fraudster. I experienced such analysis in 2016 while making a claim for a lightning strike on my home. So what has this to do with the ethics of insurance, then? Well, if an investment manager can analyze the voice of a senior executive in this way, why shouldn’t the regulator do something similar with the same people? The regulator could ask senior executives to talk about their plans, activities and achievements relating to ethical issues like integrity and fairness. Given that senior executives and key decision makers in U.K. insurance will soon be subject to new regulations that emphasize their individual accountability for ethical culture within their firms, this step would simply be taking an established practice within the sector and applying it to new ends. A lot will, of course, depend on the questions you ask. If these focus on belief and commitment, then scores could be quite high, but if they focus on actions and outcomes, then some people might struggle. And remember that U.K. insurers needn’t wait for the regulator on this. The Senior Managers and Certification Regime requires insurers to undertake their own integrity assessment of senior managers and key decision makers. Perhaps affect analysis could form part of that assessment? The results could then be used to configure personal performance plans and learning schedules. I wrote about the rise of panoptic regulation back in 2015 (link), in which regulators access and analyze real-time decision data in a continual stream from insurers. Putting artificial intelligence to use in this way would be a small but significant part of that wider development, providing regulators with critical insight into the tone from the top in a particular firm. See also: Why AI Will Eat Insurance   Perhaps the biggest signal the insurance market could take from developments like this would be that of a regulator becoming more sophisticated, prepared to get more under the skin of those they’re dealing with. Just like insurers are, some might say, in their relationships with policyholders and claimants. One word of warning, though. It is particularly important that the algorithms underlying this branch of artificial intelligence are properly trained. If that training has been carried out on the voices of the white, male executives who have largely dominated the board rooms of insurance firms to date, then this sort of AI-based analysis would turn into a barrier for the various diversity initiatives underway in the insurance sector at the moment.