Last week, the Wall Street Journal reported that Volkswagen will be testing a cloud-based quantum navigation app in Lisbon. This quantum routing app is backed by D-Wave Systems; it will use cloud-based quantum-computing services to calculate the fastest route for each of nine buses in near real time using millions of real-time data points. Since 2017, D-Wave has offered its 2,000-qubit quantum annealer (the D-Wave 2000Q System), and the company recently announced a 5,000-qubit quantum annealer to arrive mid-2020. Through Leap, its cloud-based platform, customers can get real-time access to D-Wave’s 2,000-qubit quantum processor.
Quantum annealing is best suited for optimization problems: What’s the most efficient/optimal configuration among many possible variable combinations? Earlier this year, Volkswagen conducted other experiments to optimize traffic flows in Beijing; this was run in partnership with Google and D-Wave Systems. This algorithm was successful in reducing traffic by choosing the ideal path per vehicle. Annealing applies to many industry problems; it is the least powerful and most narrowly applied form of quantum computing. Today’s classical supercomputers are just about as fast as today’s quantum computing annealing machines.
There are other flavors of quantum as well, including quantum simulations and universal quantum computing. While the hardest to build, universal quantum computing will have the most impact on insurers in the long run. Researchers have been designing algorithms for this universal quantum computer for years, and over 50 exist today. Examples include Shor’s algorithm for factoring numbers (code breaking) and Grover’s algorithm for quickly searching unstructured and massive data sets (advanced internet search). Google recently announced that, in collaboration with the US government, it has achieved quantum supremacy, solving a problem that the most powerful supercomputer in the world can’t solve.
Quantum is now moving away from pure science to practical applications. The Wall Street Journal article mentions that “by 2023, 20% of organizations, including businesses and governments, are expected to budget for quantum-computing projects, up from less than 1% in 2018.” What does that mean? First, prime number calculations used for data encryption will become exposed and vulnerable. Second, predictive analytics models and AI models, including natural language processing, will likely become much faster and more successful in generating insights.
This will trigger an arms race in insurance. The carriers that have access to these capabilities via the cloud will be able to understand and price risk better within the underwriting process, as well as adjudicate claims more effectively with less and less human intervention (i.e., lower cost). At some point, probably in the 2030s, these models and general purpose machines and robots that access them will be fully autonomous, training themselves, reprogramming, and optimizing in real time. Humans will always remain part of the underwriting and claims process, but the areas in which machines can help them will grow exponentially. We may see a future where millions of portfolio stochastic models can be run concurrently, impacting the pricing and understanding of risk pools.
The most immediate thing carriers need to worry about is security. Carriers will need to be aware of the latest developments in routing, networking, and security algorithms as they evolve in the next five years, and they should be prepared to continually update their security and network infrastructure. Like the 1970 book by Alvin and Heidi Toffler, Future Shock, people may be dealing with too much change in too short a time. As we enter the 2020s, welcome to Future Shock!
Add new comment