Most consumers and professionals of all types have a basic feeling about technological innovation as something positive. It is true that we may bemoan the loss of a favorite aspect of the past, and tend to recall for the most part only favorable situations that strengthen such memories. But, in general, people feel upbeat about the convenience and capabilities that technology can provide. We evoke pleasant feelings from the past and that is our nature. It is also our nature, at a very deep biological level, to anticipate the future.
Jeff Hawkins, in his book On Intelligence, highlights research he both collected and directs about the physiological aspect of how neurons in the brain are connected. He has shown that prediction is basically wired in to a large portion of neural circuitry. Hawkins named this approach as a "memory-prediction framework." He also makes the case for prediction as being one of the foundations of human intelligence.
The ongoing and historical basis for managing risk is directly tied to prediction as well. Being able to predict the future effects of any given risk is the basis for managing both foreseen and unforeseen events. Typically, this has meant preparing for the negative monetary aspects of adverse events whether that is an automobile accident, storm damage to a dwelling, or loss of business continuity due to a calamity. Money became the "liquid action" whereby efforts to repair, replace, or otherwise mitigate the effects of a "risk-that-turned-into-an-event" that required remuneration or restoration.
[Insured and total losses from catastrophes for the first half of 2014 are about half of an average year: Few Global CAT Events in 1H 2014: Munich Re]
Standard practice today has actuaries evaluating risks from experience as well as using models to predict outcomes, and setting rates based on mathematics and regulation. Line-of-business professionals are leveraging predictive analytics to detect patterns and wring more usable information from insurance data of all types. However, to go beyond standard practice and to prepare for the forthcoming next generation of risk management and mitigation, the innovative capabilities of cognitive computing will be required.
Cognitive computing makes use of the outcomes of research into artificial intelligence (AI) and the neuro-physiological basis of human cognition to aid computations of all types. In the strictest sense, this type of computing generally refers to a "bottoms-up" approach where processing elements (in hardware or software, but hardware is faster by far) are designed to emulate the processing of biological neural networks like those found in human brains. The cognitive approach can also make use of the more traditional, top-down (symbol-based) type of AI that we have seen via the use of knowledge-based systems (KBSs) to capture and apply our experience (via rule-based and case-based systems which are types of KBSs). In the insurance industry, we use mechanism of this sort in the form of expert systems for underwriting and claims decisioning, and other areas such as fraud detection. The elements of KBSs are also used to manage business rules to control program execution and insurance workflows.
Another essential and positive aspect of cognitive computing comes from focus on the enablement of better interactions between computers and the people that use them. The bottoms up approach that this type of information processing leverages is inherently less "brittle" and more flexible than the more familiar "black box" and expert systems. Systems and solutions that employ cognitive computing techniques can help guide business users and field professionals in tasks such as trend analysis in marketing, pricing, and sales contexts. These tools can also augment and help filter data and information to combat fraud, as well as help in the design of risk management or investment programs for policyholders.
Cognitive computing will be essential to power critical decisions that go beyond the current use of rule-based knowledge and the retroactive use of the results of predictive analytics. The aforementioned next generation of risk management (solutions that are evidence-based, personalized, actionable and much closer to real time with dynamic durations) will require big data feeds and tools that make sense of that information flow… according to my cognitive computer.
[What are some of the emerging trends in information risk management and security? Get a business and technology perspective at the Interop New York session What's Next? Emerging Trends in Information Risk Management and Security, Thursday, Oct 2.]
Jamie Bisker is a senior analyst with Aite Group, specializing in Property & Casualty Insurance. He brings to Aite Group more than 28 years of experience across insurance lines of business, having served as an IT specialist, research director, consultant, and global thought ... View Full Bio