Financial services have been paper-based businesses from day one – and the insurance sector offers prime examples of some very successful paper-based businesses.
While technology has become of growing importance in the sector for decades, slotting generative AI into such businesses looks like it could be a long process, with some easy early wins followed (slowly) by some much tougher ones.
Customer service looks like being one of the first areas of insurance business that will fall to generative AI. Some experts are predicting that almost all customer service roles in financial services will be taken over by AI in between 12 and 60 months.
This will take more than just the chatbots that we see operating today in insurance companies, but the savings in both time and money are predicted to be game-changing.
Proponents of AI suggest that fears over job losses are overblown and that the real use of AI is as a tool to help people do their jobs. Real life experience suggests that this is at best naïve and at worst disingenuous.
Insurance CEOs are focused on the bottom line and ‘extra people’ looks like rapidly becoming a luxury they cannot afford – in spite of protestations to the contrary.
APAC nations that have surfed the wave of business process outsourcing by setting up call centres at scale to provide ancillary customer services to the insurance sector could be most at risk, with the prospect of hundreds of thousands of jobs on the block virtually overnight.
The fear of job losses is one of the impediments to the swift implementation of AI in insurance but there are also worries over how the regulators in different jurisdictions will view developments. Protection is such a crucial element of societal stability that insurance regulators will want to be absolutely certain that there is human oversight of important financial decisions.
Some regulators fear that the increased use of AI will open loopholes in established systems that could make them vulnerable to hackers and other malicious actors.
There are few industry-wide studies looking at attitudes towards AI adoption in insurance in APAC, but a recent Capgemini study indicated that only 6% of retail banks were ready to use AI at scale across the enterprise – so the signs are not encouraging.
The tendency for AI to hallucinate also preys on the minds of regulators. It might be OK for an AI to make up historical events for someone writing a blog, but hallucinations within financial transactions could be catastrophic.
Another common worry for C-suite insurers is being certain that the data that they put into the AI does not find its way out into the public domain. If this is worrying insurers, it must also be making regulators fret.
But perhaps the toughest question for insurers to answer is precisely what net benefits AI will bring. Predictably, many CEOs say that it is too early to tell.
The changes brought by AI are likely to massive. We just don’t know what they will be yet. A
Paul McNamara
Editorial director
Asia Insurance Review