December 16, 2015 | Written by: Dan Briody
Categorized: Data | How To
Photo credit: Jon Simon/Feature Photo Service for IBM
The greatest early potential for cognitive computing (systems that learn at scale, reason with purpose, and interact with humans naturally) is as a decision support system. Cognitive systems will churn through massive data sets—both structured and unstructured—and offer suggestions on anything from the treatment of a disease to the risks and rewards of an acquisition. Combining the computing power of cognitive systems with human capabilities—such as judgment, empathy, and intuition—requires a new way of working. Here are a few steps to help set a proper foundation for cognitive computing in any enterprise.
Step 1. Lay the Big Data groundwork
Business leaders need to whittle insights from enormous data sets to understand and improve their strategic operations and decision-making. To do so, they need to wrap their arms around Big Data. Most organizations have traditionally relied on structured data—numbers, dates and text that fit easily into fields and can be pulled into files and spreadsheets. But, the vast majority of all data streaming in and out of companies today is unstructured (as much as 80 percent according to Gartner) consisting of images, symbols, video and natural language, little of which conforms to conventional data analysis rubrics.
A strong understanding of the Big Data both inside and outside of the organization can help business leaders analyze structured and unstructured data and glean deeper insights into the forces affecting their organization’s performance. That integration is key to making well-informed decisions that take into account all possible variables.
Step 2. Use cognitive computing to penetrate the complexity of Big Data
While Big Data can help an organization fill in the contours of some of these issues, capturing variables and identifying patterns that could point to a spike in interest rates or tapering of consumer demand, for instance, the heft and volume of Big Data analytics can be overwhelming in its own right. Cognitive computing systems can penetrate that complexity, offering business leaders the means to pose questions, interrogate the data, and fine-tune the analysis as “lay people.” They can extract the value from Big Data without having to leap through algorithmic hoops or acquire a degree in data science.
Irving Wladawsky-Berger, a former IBMer and current contributor to the Wall Street Journal’s “CIO Journal,” puts it this way: “Cognitive systems can analyze many thousands of options simultaneously, including the large number of infrequently occurring ones as well as ones that the expert has never seen before. It evaluates the probability of each potential answer to the problem, and then comes up with the most likely options.” That’s important because analytically astute organizations are 260 percent more likely to be top performers than analytic beginners.
Step 3. Examine opportunities to leverage data insight for competitive advantage
One of the new skills required to take advantage of cognitive computing is learning to ask the right questions. With technology this powerful, it’s hard to fully comprehend the extent of the capabilities. But by carefully considering the available data sets, and applying them to the strategic goals of the organization, previously intractable problems can be solved.
For example, oil and gas companies are using cognitive computing to assess potential exploration sites, taking into account the geologic attributes of a given location, but also its political climate, economic stability and weather patterns. When combined, data sets like these can provide more accurate risk/reward scenarios and significantly reduce waste in natural resource exploration.
Step 4. Be selective
Cognitive computing is a learning partnership. Because it takes upfront investment to train the cognitive system and curate the facts it needs to know, business leaders need to direct that investment toward strategically important questions where cognitive systems are likely to deliver the greatest return. Often, that’s open-ended problems with a significant amount of reasoning complexity.
Once the cognitive computing platform “learns” the language of the problem set, it can interact naturally with teams, pairing human judgment and subject matter expertise with agile and powerful analytics. The most successful implementations have come through collaboration.