Man and Machine, Teaching Each Other
AI will transform insurance, but it will be a collaborative effort
By Joel Makhluf
The rise of artificial intelligence is gaining momentum, and, coupled with data collection on an unprecedented scale, it heralds a new age of insights for industries like insurance.
Last year when a master of the strategy game Go was beaten by an artificial intelligence (AI) program, it was considered a landmark moment. There are important lessons in the story of how AI reached this level of proficiency. The AI studied past games of champion players, but it was only by playing different versions of itself millions of times that it got good enough to win against a human master.
We can only guess at the potential impact of AI on our ability to predict events in the longer term, but before we get there, humans and machines must learn from each other.
“This is augmented intelligence where we are teaching the machine, so the machine is getting smarter,” explained Dr. Anand Rao of PricewaterhouseCoopers at the 2017 Property Innovation Summit. “Then, because the machine is smarter, we are making better decisions.”
Applications for Insurance
There are many examples of where AI can add value to the insurance industry. The process is already underway in claims processing, where AI is rapidly reviewing data and documents to offer recommendations, which claims professionals can sign off on. The workload for claims professionals is dramatically reduced so they have more time to exercise their expertise, and every decision they make is teaching the AI how to do a better job.
Clearly, AI is already having a disruptive impact on the insurance industry, but we expect to see it soar in the near future. As many as 75 percent of 550 insurance executives from 31 countries said they believe AI will either significantly alter or completely transform the overall insurance industry in the next three years, according to Accenture’s Technology Vision for Insurance 2017 report, released in April.
It’s impossible to determine all the ways AI might transform the industry, but image processing is an area where it has been able to quickly add value. Using satellite images or even footage from drones, AI can analyze and categorize everything from city blocks to building sites to weather systems. It can extract insights and offer proactive advice about potential problems so those problems can be dealt with before they turn into claims events.
Employing AI in this way isn’t just efficient, it’s also much cheaper and easier than any system of human inspection. The combination of intelligence and data—especially when factoring in the Internet of Things (IoT) and all of the ancillary sensor data that can now be drawn upon—is extremely powerful. It lifts the whole process of risk assessment to the next level, boosting our ability to do everything from predicting the likelihood of floods to flagging structures for repair, simultaneously making the world a safer place and saving insurers money.
How to Employ AI
Any new technology that creates a buzz attracts attention, but in order to truly leverage the benefits, you have to consider how best to employ it.
“Don’t start from the technology,” Rao suggested during his presentation. “Start very much from the problem that you’re trying to tackle.”
In other words, consider your aims. Do you want to work on underwriting, claims, or customer engagement? Are you trying to boost satisfaction? Do you want to reduce expenses? Find the metrics that are important to your task and set out a clear objective, then you can consider how alternative data sources might enhance your goal. Next, you can move on to prototyping, teaching, and assessing your AI.
You might want to develop intelligent chat agents to guide potential customers to the right coverages or help them navigate the claims process. You may be looking to automate the collation and analysis of multiple data sources to generate recommendations for decision-makers. You may even be trying to understand what markets to enter or how to optimally price products. All of this requires a concerted period of learning and cooperation between man and machine.
“This partnership plays to the best capabilities of man and machine, with each improving the performance of the other continuously,” Rao noted.
Early adopters are already gaining key advantages. Experiential knowledge is quickly lifting AI to new heights. There is real opportunity for insurance to evolve and recapture public imagination, driving important proactive improvements in the safety and security of society through the application of AI.