In my last blog I mentioned that today we are at the beginning or the dawn of the cognitive computing era. And so, what does that mean? There are three eras of computing and each different from the one preceding it.
Let us start with the first era, which actually we consider as the tabulating era. This started maybe in the late 1800s, turn of the century, and was really about machines that were doing counting and tabulation, punch card readers or maybe special purpose machines. These helped with tasks such as controlling industrial looms or measuring voting and census data. They did it really well, but were ultimately limited in that single task.
Then came the programmable era. This starts with the dawn of the digital computer around in the 1950s. And the big change here is that you have general purpose computing systems that are programmable — they can be reprogrammed to perform different tasks and do a variety of things. People created software that used programming languages to give computers more complex tasks. Their performance, however, was limited by their adherence to established processes and decision-trees, by their inability to find relationships within unstructured information and they were also somewhat constrained in the way they interact with humans.
But what we see today and emerging over the last few years is something we call cognitive computing era. The major driver for this era is this sudden exponential increase in the amount of unstructured data that’s out there. And so the challenge is, what are the computing technologies that can really leverage all of this unstructured information in a much more natural way? And we believe that the only way we’re really going to survive with this onslaught of data is to create whats being called cognitive computing systems.
Today’s cognitive systems understand, learn, and communicate with people in natural language rather than software code. They can extract meaning from large amounts of visual, verbal, and numerical unstructured information; and they can learn as they do so, helping people make complex decisions based on Big Data. This era is leading to creation of automated IT systems that are capable of solving problems without requiring human assistance.
Lets get a taste of the cognitive systems through the means of an example. Lets consider the following question -“Did Maya shoot an elephant in her pajamas?”
There are some ambiguities in the statement – Did Maya shoot a gun or a photo? Was she or the elephant wearing pajamas? Although a human mind can resolves the question’s ambiguities based on context, but a conventional analytic software program cannot. A cognitive system is capable of resolving these ambiguities. First it creates multiple hypotheses about the meaning of question elements such as what object was shooting and who was wearing pajamas. Then it examines those elements against the context of its corpus–the set of documents that constitutes its essential knowledge. And thus it understands the most likely meaning of the question. Then it can answer the question providing a measure of its confidence in that answer.