Cognitive Computing and BPM

A major transition is shaping up that will change BPM in radical ways that are hard to imagine.  In the past, when most of us thought of process work, we depended on models derived from work done in manufacturing organizations in the 70s and 80s.  The idea of a process flow was often derived from the auto production line, with clear, linear steps, specific activities at each step,  all resulting in a specific product being the output at the end of the process.

For some time now, the dominant process model has changed, and most organizations are spending more time analyzing and redesigning processes that involved more-or-less continuous customer involvement.   Imagine, instead of an auto production line, a customer interacting with a bank, opening accounts, cashing checks, making deposits, calling to inquire about statement discrepancies, applying for a loan, and so forth.  These ongoing processes don't have a specific product, but are made up of a more-or-less continuous series of customer-bank interactions, each critical to the overall success of the process.

Recently there is a new emphasis on even more dynamic processes — or on capturing aspects of service processes that had hitherto been ignored.  Thus, today, we are as likely to concern ourselves with how bank employees make decisions as with how they  physically interact with customers.  For awhile, this new concern seemed to fit under the cover of Business Rules, but it is rapidly evolving well beyond that.  Some now prefer the term Case Management and others are speaking of iBPM (intelligent BPM).  Even these terms are probably inadequate to capture the scope of the coming change, which also involves analytics and big data.

IBM has recently coined the term Cognitive Computing, and the CEO of IBM has been quoted as saying that “Cognitive computing is the future of IBM.”  The term Cognitive Computing has often been associated with Watson and with Watson's recent success at Jeopardy!, a popular TV game show that involves answering questions about a wide variety of popular lore.  In a recent contest with several past winners of Jeopardy!, Watson demonstrated that it could easily beat the best human contestants.  In essence, Watson received the question (it was input in typed form) when the human contestants did, then signaled when it had the answer and gave a verbal response.  Thus, in essence, Watson searched massive databases, prioritized possible answers, and then rendered its preferred answer into spoken language.

For someone like me, who was active in expert systems development in the Eighties, this is a rerun of the promise of artificial intelligence, with much enhanced capabilities.  Clearly we are about to witness another major advance in what we expect from computers.

As someone active, today, in BPM, it suggests that we will soon have the ability to automate business processes that deal with large amounts of data, complex decisions (decisions where we choose the best of several possible answers), and spoken human interfaces.  (I'll leave aside the advances that have occurred in robotics, but clearly much of this new technology is going to be embedded in devices, from smartphones to cars.)

Ask yourself, as a business process person, how you plan to select appropriate activities to be automated by these techniques, how you will analyze the activities, capturing the knowledge, and the decision rules needed for both the decisions and the verbal interactions, and how you will represent those facts and the process flow.  I suggest you probably won't do it with BPMN, as currently configured.  At a minimum, you will need some way of capturing concepts in a cognitive network map.  And you will need a way of capturing probabilities to be applied in uncertain decisions.

Recent OMG efforts to create standards for Case Management and for Decision Management are just the tip of the iceberg.  We are going to need to explore a lot of complex analysis technology to come up with good ways of understanding the world of process as it will exist by the end of the next decade.  This effort is just beginning.  We can all prepare for it by trying to be more flexible in how we approach BPM.  The old certainties will begin to fade as new techniques become available, and we are all going to have to struggle to incorporate new ways of thinking about processes in our methodologies.


Speak Your Mind


This site uses Akismet to reduce spam. Learn how your comment data is processed.