December 14, 2016 | Written by: Marie Glenn
Categorized: Cognitive Enterprise
Share this post:
Risk managers in every organization are deluged with information. A considerable portion of their day goes to tracking down data of one kind or another, parsing and assessing that which they find, comparing it against other information, and gauging the relevancy and impact in order to make a judgment. There are screening and vetting tools to help, of course, databases and information engines intended to separate the wheat from the chaff and surface the most pertinent findings. But few of those tools are capable of acting with the nimbleness, reach and scale demanded of today’s business environment.
As the number of information sources grows and the formats and channels expand beyond text to sound, voice, images and other unstructured data, the need for robust mining and modeling has grown especially acute.
That’s why cognitive computing, machine learning and natural language processing have the potential to be transformative. According to IBM research on cognitive early adopters, nearly half of financial institutions say they increase market agility through their cognitive initiatives while reducing risk. The ability to tease out patterns and leading indicators, run that data through a volley of scenarios, and present key correlations and insights allows analysts across industry to apply their knowledge more effectively—and address scenarios and sample sizes whose complexity and scale can exceed the limits of human cognition.
From a risk management point of view, an evidence-based information filter can make all the difference. Cognitive platforms generally provide not just one answer, but several. Each opinion, recommendation or finding comes with a confidence interval. That interval is determined on the basis of the quality and reliability of the sources used, and the level of faith the system has formed after repeated rounds of testing and learning. As Marc Teerlink, Chief Business Strategist for IBM, noted, “Cognitive adoption allows for a different way of governance, a different way to play with numbers, and much greater space for scenario thinking.”
Cognitive analytics can allow credit and risk managers, insurance adjusters, auditors, investment analysts and others to become both more productive and more predictive. Less time chasing data and number crunching means more time helping customers and clients apply the insights.
To get the most out of cognitive as a high-functioning filter, however, it’s important to pick the right business scenario. Look for an issue that affects a critical mass of customers and has enough data to support it. If the program or question is too narrow in scope, the investment will benefit only a narrow slice of the business. It’s also important to back the program with the right level of resources. Training cognitive systems takes time, an openness to trial and error, and the ability to work with relevant user groups to refine the business rules, semantics and other inputs needed.
Applied well, cognitive tools can help risk leaders punch through the noise of today’s information environment to access and employ the insights that matter most.