The survey of almost 200 insurance executives showed wide adoption of machine learning and suggests it will bring “significant” change to the industry over the next three to five years, Earnix said.
Of the firms using the technology, 70% are doing so for risk modelling purposes. More than a third of the insurers that have adopted machine learning technology use it for demand models (34%) and fraud detection (36%).
Most of the firms using machine learning said they had already seen measurable benefits, according to Earnix. Over half said it has made their analytical models more accurate, leading to better risk assessment and better decisions, it said.
According to the survey, the main barrier to adoption is a lack of expertise within firms, with 82% saying they are relatively inexperienced with machine learning.
Udi Ziv, chief executive of Earnix, said: "The insurance industry has always been data-driven and the use of machine learning is a powerful trend that will quickly become a competitive edge for those who embrace it."
Luke Scanlon, a fintech legal expert at Pinsent Masons, the law firm behind Out-Law.com, said: "The more insurers can use machine learning and other artificial intelligence tools that help in assessing and pricing risk, the more they will be able to service their customers effectively."
"However, both regulatory and non-regulatory barriers can make this difficult – from concerns around compliance with the principle of data minimisation, to restrictions on the use of health data, to dealing with legacy systems and difficulties in accessing data held by third parties. As regulators continue to assess the use of new technologies used in insurance, they must ensure that further barriers are not arbitrarily put in place which would make it more onerous for insurers to innovate," Scanlon said.
The Financial Stability Board said in November 2016 that it would scrutinise the way that artificial intelligence could affect financial services.