While there are many definitions of knowledge management and many things it governs, there is a slow-moving train approaching we can see from a mile away that we seem to deny its existence. All aboard the artificial intelligence railway–either buy a ticket or hang out at the station. Artificial intelligence and cognitive computing are here to stay and will become the biggest disrupters this generation will ever see in their lifetime. AI and CC will disrupt the very way in which doctors, lawyers, accountants, and yes, knowledge management professionals practice their trade.
To put this discussion in a bit of context and for purposes of this article, traditional knowledge management applies to the classification of documents, data and people, to name a few of the pieces, into a taxonomy that allows easy recall of information when requested. Historically, this has been a very manual process involving human intervention and thought when organizing bits of information in the hierarchical taxonomy structure. Additionally, the humans involved were typically former practitioners of those they are organizing the information. This typically was a very skilled person and not cheap labor by any means.
Thinking about the statement above regarding artificial intelligence and cognitive computing, context should be placed around these technologies as they are often used together but synonymous in meaning. While, more often than not these two terms go hand-in-glove with each other they are not the same.
Quoted from Peter Novig, research faculty member at USC Berkeley and Director of Research at Google, “artificial intelligence is exhibited by intelligent agents that can decide what actions to take and when to take them.” Using this definition in relation to a Tesla automobile, the Tesla is pre-programmed with millions of inputs that the car can use for hands free driving. The car cannot learn anything more than its last firmware update whether from the factory or an update. While the car itself is quite smart, it lacks the capability to “learn” new things or self-correct its programming.
Cognitive computing, however, is the mechanism whereby a device such as the Tesla could “learn” new things about new driving conditions for example. IBM defines cognitive computing as “…systems that learn at scale, reason with purpose and interact with humans naturally”. Cognitive computing has the ability “learn” or extend its base programming with which it started. In other words, if it doesn’t know an answer or response to a question it can ask what the answer or response is for the next time it is requested. Using the Tesla again as the example, Tesla’s are programmed extensively with road conditions, lane design, traffic patterns and typical vehicles it may encounter.
If, perhaps, it encounters a flying drone in the lane the Tesla is driving it may not know what to do as its “artificial intelligence” side of the equation never anticipated a vehicle that had nothing underneath its body. If the Tesla had cognitive computing built into it, the Tesla could ask the driver, “I see something floating in front of us–should I treat this is as if it were normal vehicle to be avoided”, requesting the driver to respond so that it can learn for now and the future what to do when it encounters this situation. Additionally, this type of learning could pass this “learning” back to Tesla headquarters to be pushed out to the other Tesla’s driving around to effectively crowd-source new learning abilities.
One day artificial intelligence and cognitive learning will supplant traditional knowledge management
Considering these two technologies, their application to knowledge management should seem applicable. For example, a computer could “learn” what documents are made of and use this learning to file and classify them. If you look at the makeup of a letter, there is a date at the top, an address block, the body and a signature block. A computer examining document files on a hard drive could take what it knows from what it has been taught and determine which documents are letters. Additionally, the computer could extract metadata from the document or perhaps pull out bits of the contents to learn what the letter is about to build a database with this information. Taken a step further, a search engine could be used to query this database, but because it too can use inputs from what it knows about the person querying the database it can rank the results in accordance with metadata about the person running the searches. As an example, a lawyer in the real estate practice group may do a search and the search engine may already know that the attorney is in the real estate group, the context of last searches they’ve done and last documents they’ve worked on dealt with specifically real estate divestitures. The search engine would use this information to narrow or rank the result based on this new information. From a cognitive standpoint, if the attorney changes their practice to the mergers and acquisition group the search engine would learn this based on the new types of searching that person is doing. All of this without human intervention or mechanical categorization of documents.
Have you been to Amazon’s website, searched and filtered results to find what you are looking for? That’s it. Have you later then traversed to Facebook and seen pop-ups of what you were looking for at Amazon? That’s loosely it but all the same. Have you looked at airline tickets and closed-down the site because of the price only later to receive an email from the airline for a slightly lesser price? That’s it. While these examples are marketing related, these will shortly be translated for business purposes. For example, assume you are working on a business-related task such as writing a new company policy. You may first want to search for past or like policies but when searching notice that the search engine you are using quickly bubbles these to the top of the results like Google does. Our computers and devices (a la IoT) are constantly gathering information about what we do, what we search, where we go, our timing, etc… These inputs are gathered in a way that propose to help us in our busy lives and for business purposes.
Will traditional knowledge manage be forever transformed with the advent of artificial intelligence? One day artificial intelligence and cognitive learning will supplant traditional knowledge management and arguably has already begun. Fear not my KM friends, like attorneys who think they will all lose their jobs because of this, it still needs human input- for the time being. These technologies are great but without human intervention, training and designing purposeful tasks for them they are just computers. Like anything more advanced than humans we tend to fear possible scenarios of human takeover by machines, but if we are smart we will embrace these technology changes in the same way we embrace Facebook upgrades–they just happen.