The latest edition of WTW’s Radar pricing software has been released. According to the company, Radar 4.15 introduces a “market-first” capability that allows insurers to benefit from machine learning’s predictive power without sacrificing the ability to accurately interpret the outcomes of increasingly complex models.
“The innovation that underpins Radar 4.15 is ground-breaking,” said Serhat Guven, managing director at WTW. “For the first time, insurers will have access to state-of-the-art, speed-to-accuracy performance that goes hand-in-hand with unequalled levels of transparency.” “While technological advancements will certainly continue to transform the insurance sector, harnessing the full potential of new technologies will only be achievable if customers and regulators trust insurers to utilise them responsibly.”
AI-based and machine-learning tools offer better predictive performance and faster time to market, but they are also more sophisticated and opaque, preventing users from completely comprehending the model they are constructing, according to WTW. If used without prudence, this “black box” nature can pose considerable concerns, including magnifying bias risks that lead to biassed judgements. WTW reports that regulators are quickly losing patience with apparent unfair pricing.
“Radar 4.15’s powerful new algorithm addresses machine learning’s black-box limitations by providing insurers with a ‘best of both worlds’ solution that combines the pure predictive power of gradient boosting with the exceptional interpretability and transparency of generalised linear models,” Guven said.
Radar 4.15 also includes improvements to Radar Base and Radar Workbench.