This article first appeared in the program of the International Cargo Insurance Conference 2018.
Like it or not, Artificial Intelligence (AI) is here to stay. Insurance is data heavy and AI has the ability to access and analyze it in a way humans cannot. By leveraging such technology to process massive amounts of data quickly, insurers can automate processes, reducing both time and costs.
However, it also brings new risks. A report by PwC in 2017 stated that some 30% of existing UK jobs could be at risk from AI by the mid-2020s-2030s. It is therefore clear the marine market has a duty to understand both the advances and challenges that lie ahead.
Companies worldwide such as Tesla, Amazon and Microsoft are embracing AI technology and the insurance sector is also increasingly exposed to its capabilities. The question, however, is whether it will simply disrupt or fundamentally alter our industry?
In all its forms
AI is the “simulation of human intelligence processes by machines, especially computer systems.” It comes in two forms: narrow — the every-day technology that powers virtual assistants or autonomous vehicles; and general — where AI more closely mimics the adaptable intellect found in humans.
One key AI-based approach is machine learning, a process whereby the more a computer is fed information, the more accurate its predictions become. In the personal injury arena, insurers are looking to use it to automatically assess personal injury claims and predict repair costs from historical data.
In the motor insurance industry, Direct Line has partnered with Tesla to launch “Insure My Tesla,” offering a premium discount to Tesla drivers based on the fact that its autopilot vehicle scheme has reduced crashes by 40%. The quid pro quo for the insurer is access to the Tesla records providing real insight into the use of autopilot on the roads.
However, the lack of control over data collection may be problematic, especially in light of recent headlines around social media data and the recent implementation of GDPR legislation.
Internet of Things
Analysts forecast that over 50 billion objects will be connected to the internet by 2020, making it fertile ground for logistics operations. Telematics (a form of IoT) is already used in the motor insurance industry to provide discounts for safe driving, low mileage or frequency of journeys. It is also envisaged that AI and machine learning will help minimize the risk of accidents and reduce claims.
Combining machine learning and IoT has the potential to create a strong competitive advantage for smaller insurers specializing in niche areas, providing access to deeper analytics, which in turn spurs innovation, reduced costs and facilitates better risk management.
Trials are ongoing for almost every mode of autonomous transport, from self-drive fleets such as Waymo, a Google- owned company in the U.S., to the first commercial autonomous vessel, “Yara Birkeland,” due to be operational by 2020. It is also predicted that the self-driving haulage industry will take over 1.7 million jobs worldwide over the next decade.
However, whilst autonomous vessels herald a new era of shipping, existing maritime law is not set up to handle them. Similarly, how will product liability affect the traditional maritime liability and insurance rules? Will haulers be required to hold valid driving licenses, and who will be responsible if an accident does occur? Clearly, further discussions are required.
Drone technology is also evolving. Amazon has been conducting drone delivery trials in the UK and U.S. for some time now; while in China, the first permit to deliver packages using drones has already been granted.
In the insurance market, the recent U.S. hurricanes and wildfires highlighted the benefit of drones, allowing loss adjusters virtual access to inaccessible terrain and providing swift claims responses to insurers. However, these advances bring their own risks, such as regulatory issues, vulnerability to cyberattacks and privacy infringement.
The insurance industry
Many believe insurance is ripe for AI disruption. The sector is data rich and AI systems significantly improve our ability to tap this huge data well. Some are already looking at AI to evaluate loss and other underwriting information, supporting risk assessments and premium calculations. It can also help collate loss, financial and other company information as part of underwriting submissions, evaluate loss history, and ultimately determine limits and coverage types.
However, it raises fundamental questions. How do we demonstrate the analysis that led to AI-based conclusions? Claims adjusting is both art and science, so could indemnity spend feasibly go up and quality down? Where does responsibility lie for poor risk selection if AI is involved? It is therefore clear that if future decisions are to be made using AI, then the processes involved in arriving at those decisions need to be open to scrutiny.
Further, English insurance law was never drafted to interpret what facts and matters were within the knowledge of an AI system. Thus, one can see issues arising if ultimately it is the system and not an individual making decisions that affect insurance coverage. How can an objective standard be applied if every insurer’s AI system is unique, and how would a prudent AI system act?
It is clear AI creates both opportunity and risk. However, are we as an industry ready for the changes it will bring? Ultimately, the market will need to consider how it adapts and evolves to stay competitive – if we do, then AI presents us with real potential advantages.