KEY TAKEAWAYS:
--As the language models improve, the ability to reduce reliance on call centers may be coming sooner than later.
--The mundane work of auto filling applications, claim forms, coverage certificates, renewal correspondence or really any repetitive and predictable task is something ready built for an AI. AI would also be very capable at comparing coverage and policy language quickly. As the technology evolves, AI could quickly move into writing briefs and coverage opinions.
--AI tools will be force multipliers to make all work faster and more efficient.
----------
Whether artificial intelligence will help the insurance industry work smarter, or whether it will mean massive job losses, or maybe represent something in between is yet to be seen. But what is for certain is that the dawn of artificial intelligence has already come and that nearly every facet of the insurance industry will soon be reckoning with what it means and how it will fit in its future.
Large language models, such as ChatGPT, and image-generation AI, such as DALL-E, have wowed audiences over the past few months, but in many respects, machine learning and algorithms have been playing a role in many aspects of the insurance industry for years already.
Simple chatbots on websites and many underwriting tools are already using many of the baseline tools found in the new artificial intelligence tools splashing the headlines, but what is remarkable is the speed with which many of these tools are evolving and the potential many of these seem to have for quickly jumping into innovative aspects of the industry that have not yet seen AI’s influence.
“I think you are going to start to see CEOs who are hired for their ability to use AI in the very near future,” said Bill Holden, senior vice president of executive perils for Liberty Company Insurance Brokers. “I don’t know if they are asking candidates about it now, but in back of their minds I know that all the boards are thinking about it.”
Customer-Facing
With the rapid evolution of the large language models, the obvious first line of potential for AI’s application in the insurance industry is with the point of contact with the customer.
Updated web or app chatbots are certainly on the horizon, as are more intuitive phone chatbots that can move beyond simple call routing operators and move more into the realm of solving customer service problems and answering coverage questions. Front desk receptionist robots could even conceivably replace humans, presuming there is still a role for bricks and mortar offices in that future.
But anyone who has spent time dealing with an automated customer service agent could be forgiven for casting doubt on whether AI will completely replace the human touch in customer-facing roles. Still, as the language models improve, the ability to reduce reliance on call centers may be coming sooner than later.
See also: Lessons Learned on Insurance Apps
Paperwork Heroes
Moving away from the immediate customer-facing role, AI could very easily slip into an effective role just behind the scenes helping customers, and really anyone who needs to spend any time with paperwork.
AI doesn’t get fatigued, so the mundane work of auto filling applications, claim forms, coverage certificates, renewal correspondence or really any repetitive and predictable task is something ready built for an AI.
AI would also be very capable at comparing coverage and policy language quickly.
Coupling the large language models with image recognition could also allow the technology to auto populate things like claims information based on uploaded photos and help underwriters make initial coverage decisions and claims settlements based on a trove of automatically generated data points. AI can easily and instantaneously interface with sensors and images and data in ways humans just can’t.
Imagine an AI assistant assessing damages for every policyholder in a community post-disaster based on drone-captured, satellite-downloaded and customer-uploaded photography, all in a fraction of the time it would have taken a team of humans with boots on the ground.
As the technology evolves, AI could quickly move into writing briefs and coverage opinions.
And the ultimate use case would be using AI coupled with predictive analytics to prevent claims in the first place, and then taking it a step further and using it in a fraud detection role by analyzing patterns in claims data and applications that might have otherwise slipped past humans.
Bias and Discrimination
AI and machine learning can move faster than humans, but unfortunately it is impossible to see inside their black box to see what is driving their decisions. Once they are trained on their data sets, they make their decisions independently, which is their strength, but when it comes to questions of bias and discrimination, potentially also a major weakness.
In insurance, bias and discrimination are obviously illegal, but without knowing what is driving the decisions made by AI, there is the potential to amplify implicit bias that is already in the data that the AI could potentially exacerbate.
“AI just doubles down on what it thinks it knows,” said Bob Gaydos, CEO of Pendella Technologies.
With the potential for harmful assumptions to get amplified if AI gets involved in underwriting, regulators will likely take a close look at any automation that has even a whiff of potential for bias and discrimination to be introduced.
“Bring in AI,[and] it is going to be questioned,” Gaydos said. “If we open that door, we have to be ready for that discussion.”
Gaydos warns that today’s laws are insufficient to regulate AI underwriting, and a new age of artificial intelligence is likely to usher in an intense wave of political and regulatory scrutiny that the industry may not be ready for or anticipating as it embraces AI.
See also: 20 Issues to Watch in 2023
The Future
There is no doubt that AI is in the door in the insurance world.
While technically there could be the potential to remove humans from every insurance process, agents, assessors and underwriters are a long way from being replaced by the current generation of AI. More likely AI job losses will be felt most acutely with the front-line workers doing tedious work — work that had previously been outsourced to call centers already. And with the more advanced work, AI tools will be force multipliers to make the work faster and more efficient.
Now, what will the market look like decades from now? Perhaps an AI analyst will be able to give us an assessment.
