‘AI’ Disruption Could Run Far Deeper than Underwriting
May 19, 2017 by Jay Cooper
Insurers know artificial intelligence and big data could disrupt industry practices such as underwriting. But disruption could be far more profound: Less risky consumers could ultimately use machine learning to determine whether they even need insurance. The implication is one of the several large-scale impacts AI could have on the insurance industry, the Financial Times reports.
Machines could one day give customers tools to decide whether insurance is necessary. This could influence purchase decisions or, at a minimum, lead to reductions in the amount of insurance purchased. The article cites a start-up called Lapetus which aims to predict an individual’s life expectancy based on ‘selfies’.
Companies could also use AI to build their own predictive and preventive risk models, deciding what risks to keep on the balance sheets and which risks to transfer to insurers.
The article also sheds light on just how disruptive big data and artificial intelligence could be to traditional underwriting methods. It cites an Oxford University study asking machine learning experts to assess the tasks involved in 702 different professions. It ranks insurance underwriting as the fifth most susceptible to automation.
Don’t expect underwriting jobs to disappear tomorrow, however. Humans are still needed to train algorithms, deciding what data to give to machines and what constraints to give it, experts tell the Financial Times.
Regulation could also slow change. Insurers will still have to explain insurance pricing to customers, not just say that a computer determined the price.
Artificial intelligence could eventually create an entirely different model for assessing risk, leading to situations where insurers create unique policies for each individual person, the Financial Times says.
Investors are starting to take note of the effects artificial intelligence could have on the industry.
Insurtech investment tied to artificial intelligence themes totaled more than $500 million in 2016, according to Accenture.
