As our panel of experts got stuck into the latest InsurTech webinar, "Getting practical with artificial intelligence in insurance", you asked us your burning questions. In this installment of this AI series, Mark Andrews, Domain Director, Altus reveals his answers.
With Samsung showcasing Bixby and Google release their Home AI, where should Insurers focus their AI investments specifically towards?
Claims will be the area most likely to get focus from traditional insurers needing a solid business case for investment in AI. However, a lot of the InsurTech investment is in the digital brokerage space, where chatbot interfaces are looking to engage a new kind of customer, historically not interested in speaking directly with insurance companies.
Do you have any concerns with Robo advice becoming depersonalised and as a result potentially leading to poor customer service? We already see automated recordings which can frustrate customers because the options they seek are not available, ultimately leading the customer speaking to a human anyway. Is this a concern?
Pure digital-only advice (today) can provide a low cost, streamlined digital solution to meet relatively simple needs; more complex needs currently require a hybrid or human-only solution. And similarly, we have already witnessed instances where poorly developed and deployed chat bot type capabilities have left customers feeling frustrated and annoyed, rather than getting the answers they want. The European Supervisory Authorities’ (ESAs) ‘Joint Committee Discussion Paper on automation in financial advice’ identified these challenges (amongst others), and organisations are working hard to address them. In time, we will see increasingly sophisticated automated services, and consumers will only speak to humans (rather than Alexa!) to answer more complex questions.
How critical is Machine Learning, an important part of AI, for InsurTech?
It will be fundamental in the long term. Insurance is about dealing with large sets of data, whether that be for pricing, claims, fraud, customer acquisition etc. New technology and ready access to cheaper infrastructure has meant insurers are sitting on even more data sets, but knowing how to get value the data will be the challenge. Machine Learning techniques certainly have a big part to play.
What do you think the challenges will be in terms of implementing AI for bigger businesses? Will it be more about employees embracing the technology, customers trust the processes etc?
Insurance companies are by their very nature risk averse so the challenge for the larger insurers will be adoption of any new technologies into their existing operating model.
How quickly will companies that chose not to invest in updating their technology infrastructure now fall behind in the future?
The insurance industry is so slow moving that I don’t see a large proportion of incumbents being left behind from the leaders in this space, or early adopters stealing a march on their competitors. A fast follower approach for the more cautious insurer could be a savvy strategy as the investment from early adopters take time to show a real return.
Are we in a hype cycle with lots of companies claiming to do AI that are not?
The webinar discussed the practical uses of AI which referred to process automation and robotics programmes replacing human tasks with machines or technology. More advanced AI techniques are still very immature in terms of adoption. The InsurTech players are the ones likely to introduce the game changers and insurers will then need to work out how to implement the technology.
Does leveraging external AI solutions within enterprises via API's actually work?
Outsourcing AI via an API is already working for many innovative providers who have seen a business case that can leverage AI, but don’t have the resources to do it themselves. These range from small Fintech’s to some of the larger business process outsourcers using AI to improve back-office and chat solutions. The big players all have very stable and configurable AI offerings with good integration that allow you to be up and running quickly and put through good volumes. Key to success is to use that capability efficiently and ensure compliance when it is used, especially in FS.