Making machines think
Analytics is moving toward a new era. Businesses want to use machines as a cognitive platform driven by data. The term “cognitive computing” is rising in popularity, the notion being that it brings new value to big data by combining artificial intelligence and machine learning.
Machine cognition is much like that of humans. As a child grows older, his or her cognitive processes become sharper and more expansive. The same can be said for machines. Machines will, in time, produce results based on the rules, deterministic and probabilistic models—the advantage being that machines have infinite storage capacity, unlike humans.
Big data is the basis and the driving factor for the cognitive computing generation of analytics. It goes beyond processing of big data. It’s about connecting the dots of insight and delivering the right answer at the right time and in the right context (Download chart 1). As industries focus on consuming more and more structured and unstructured data together, decision-making is evolving into the realm of cognitive computing.
Cognitive systems will also help relations between humans and machines evolve. As humans escalate their adoption and use of technology, machines will gradually supplant mundane human tasks. And with humans and machines working alongside each other, machines will better understand human interactions, nuances and the environmental context. (Download chart 2) So while humans have the know-how about the machines, through their interactions, both are expected to aid each other and jointly become better—and more powerful. Imagine a customer walking into a bank and having the exact same business experience every time he or she enters because it’s run by a cognitive system—the experience is consistent, precise and appropriate, irrespective of the staff attending to the customer.
So what do organizations want from cognitive systems? Well, a few things. They want to be:
- smarter—having smarter algorithms and algorithms that innovate and continue to get better;
- faster—not just processing information but producing the analytical insights more akin to human reaction time; and
- easier—systems that are easy to work and that should be capable of understanding human language and learning from human inputs.
Big data, machine learning, natural language processing (NLP), Internet of Things (IoT) and prescriptive analytics are all components required for a cognitive system. Massive parallel execution capabilities of the machines along with highly available systems and commodity hardware are all factors that favor the development of cognitive computing. Cognitive systems compute and learn, enabling industries to move beyond rule-based reasoning to model learning, indicative of cognitive systems.
Cognitive learning is cyclical. It begins with machines recognizing and processing human language. It leverages expansive processing capabilities, along with data from various sources (like event streams, stored repositories, etc.), analyzed together to generate the probable outputs. Those probabilistic results are then shared with the user (whether a human or machine), backed by the evidence from the analysis. Based on the user actions, the machine registers and learns from each and every interaction.
Businesses desire systems to help make the customer experience unique, meaningful and seamless. Cognitive computing is poised to change the game.