Technology
A Key to Success

• Mark Breading
• December 2020
• 

Discussing artificial intelligence in insurance is like discussing digital transformation. Everyone recognizes that it is vitally important, but what exactly is it?

Both AI and digital are terms that encompass a broad range of technologies, projects and use cases. The term AI can be confusing since it represents several different technologies that range from simple to complex. Added to that is the fact that AI has been around for half a century and has a checkered history.

Insurer AI initiatives in the 1980s and 1990s that leveraged technology like case-based reasoning and neural networks had mixed results. But in the past decade, AI technologies have taken a great leap forward. The combination of massive computing power, big data and algorithmic breakthroughs have resulted in accelerated progress across the whole AI family.

For property/casualty insurers, the question becomes how to identify the business use cases that will deliver the most value today and have the most potential to transform the business over time.

The first way to answer that question is to examine the ultimate purpose of using AI. Is the objective to automate, gain new insights or some blend of the two? If the goal is to improve efficiencies and/or reduce human involvement through automation, then there are AI technologies that do that well. AI-powered process mining tools can identify tasks that can be eliminated, reworked or even accomplished via robotic process automation.

Automating interactions can benefit from chatbots to handle the most common questions and interactions. The chatbots may be text-based, in which case natural language understanding/generation technology is used, or voice-based, which requires speech recognition.

If AI is meant to gain new insights and augment human expertise, then deep learning technologies apply. Machine learning, natural language processing and computer visioning have great potential in insurance to address important problems. Not that there is an artificial dividing line between automation-oriented AI and insight-oriented AI, but there is crossover.

For example, speech recognition requires deep learning technologies, and computer visioning can extract information to improve automation. However, the technologies that offer new insights are typically the ones that provide a significant competitive advantage. Underwriting and claims are two areas where these AI technologies offer the most potential, but there are strong use cases across the value chain.

Recent Strategy Meets Action research identified the use cases for underwriting and claims with the highest transformation potential. Insurance executives reviewed 10 functional areas of underwriting and risk management. For both personal and commercial lines, they selected risk assessment as the top area. Personal lines insurers ranked inspections (via aerial imagery/machine learning), exposure analysis and submissions as three other areas with high potential.

For commercial lines, exposure analysis ranked second highest in transformation potential, followed by triage and submissions. In claims, the clear winner across all lines was fraud. There were several other areas where high value is expected, including predictive reserving, triage/assignment, first notice of loss/first report of injury and damage estimation.

There is no lack of opportunity for harnessing AI technologies to benefit P/C insurers. The challenge is in mapping business strategies and plans to the highest-value, AI-based solutions to deliver on the promise of AI for business transformation.

Best’s Review contributor Mark Breading is a partner and the chief research officer at strategic advisory firm Strategy Meets Action. He can be reached at mbreading@strategymeetsaction.com.