- Emergency Consultation Services
- Risk Management Services
- Who We Are
- Our People
- What We Do
- Why We Are Different
- What’s New
- Where We Are
By: Seth Kirby
In 1969 the United States successfully landed a manned spacecraft on the moon. Upon stepping on the lunar surface, Neil Armstrong commented on the momentous occasion by proclaiming that it represented a “giant leap for mankind.” It was an amazing accomplishment for the United States in the space race, and in many respects, the lunar missions have fueled the technological advances that we enjoy today. Consider for a moment that the computers that were on board the spacecraft used in the Apollo missions had far less processing power than the cell phone in your pocket. Indeed, the Apollo Guidance Computer that was used to control the flight of the spacecraft was more basic than most modern toasters. The entire network of mainframe computers used to monitor and control the lunar missions operated on less memory than the capacity of today’s usb flash drives. Technology has advanced at an exponential rate.
Last week, the world may have witnessed another technological breakthrough. A computer system known as AlphaGo won a Go match against 18 time world champion Go player Lee Se-dol. AlphaGo won the match by winning the first 3 games of the 5 game series, only to suffer its first loss to its human opponent over the weekend. The final game of the series is scheduled to be played today.
Alpha Go is an artificial intelligence (“AI”) program developed by Google’s DeepMind division. While other AI programs have previously beaten world class competitors in games like checkers and chess, this is the first time that AI has successfully played Go at a world class level. The significance of this accomplishment can only be understood with reference to the game itself. Go is a 2,500 year old game that is played with colored stones on a game board with a 19 by 19 grid. Unlike chess, that has a possible 35 moves in any particular turn, Go has a possible 250 moves per turn. It is claimed that there are more possible positions on a Go board than atoms in the universe. As a result, the game cannot be played at a high level simply by analyzing all possible moves, instead it is a game of strategy combined with human intuition. Thus, AlphaGo has been able to demonstrate that AI is capable of “acting” like a human including predicting its opponents behavior.
AlphaGo was created for the purpose of exploring and further developing AI. Like most modern AI systems, it is a neural network that is able to collect and analyze large volumes of data and make predictions about what will happen next based upon previous outcomes in similar situations. It has improved its play by first studying millions of human moves and then using those game models to play other versions of itself, constantly looking for which moves will generate better outcomes in the context of the game board. AlphaGo is not just playing a move that has the highest chance of success, it is also deciding upon that move based upon its opposition’s predicted response. It is not hard to imagine how this improved strategic thinking and learning could have a significant impact in the insurance industry.
The insurance industry is based upon the concept of spreading the risk of loss among a group of policyholders. The careful collection and analysis of data by an insurance company allows the company to anticipate how often a group of similar individuals will experience a loss and the likely monetary value of any loss that occurs. This underwriting process enables the carrier to create and price policies so that the anticipated losses are covered and the carrier is able to generate a profit from the endeavor.
AI could improve this analytical process so that risk analysis and underwriting is based upon far more than basic demographic information. Instead, AI could be used to examine an applicant pool’s insurance profile and other available information (Facebook posts, cell phone usage, hobbies, educational background, etc.) to more accurately determine their risk of loss based upon the likely future conduct of the group. Not just developing a risk assessment, but actually predicting their losses.
Before you dismiss this thought as sci-fi fantasy (a la Minority Report), you should know that Amazon already engages in this type of predictive analysis for its Amazon Prime customers by shipping products to local distribution centers for delivery before they are ordered by their customers. With further development, AI has the potential to become a game changer for the insurance industry. Only time will tell.