CHATBOTS: The digitalisation of knowledge through a new lens

79% of executives now believe seamless and digital access to knowledge is very important to overall business performance. Executives understand that there is too much dependency upon knowledge in people’s heads. And, even when it is documented, there is inadequate transparency and traceability to the way it is used to make decisions, when applied in practice.  

Often, the best starting point for digital transformation of knowledge is when it is documented. Knowledge contained in documents consist of seemingly endless pages of content, which is typically not easy to use. This is not surprising, as it is frequently left to the user to synthesise the knowledge in context to their needs. This fact alone has led to systemic weaknesses that are hidden in plain sight and has become the causality of so many executives grappling with frequent unexpected problems.  

Early attempts at knowledge management led to an overhyped market followed by a deep trough of disillusionment. So much has changed. The digital revolution has already upended and transformed so many industries. As knowledge continues to grow in complexity, it is being matched by a growing momentum to think about knowledge through a new lens, so that it becomes a measurable and working digital asset.

There is no doubt that significant breakthroughs for the digital access to knowledge have already occurred, driven by data and artificial intelligence. This includes the rapidly growing use of machine learning, deep learning, robotic process automation, virtual reality, virtual assistants and chatbots using natural language processing. Even with these advances, it has become better understood that these capabilities have both strengths and weaknesses.  

One area digitalisation has made little headway, is knowledge represented in hundreds of millions and perhaps billions of rules embedded in endless pages of content, found in so many documents. This type of knowledge has always been human controlled and covers subjects such as regulations, statutes, trade agreements, tax, policies, procedures, standards, manuals and instructions.

It is simply untenable for machine learning to upend and change these rules. Imagine reporting to the Board of a private or public organisation that deployed machine learning has been empowered to change policies and regulations where it deems appropriate. This scenario is not only farcical, it draws a clear line between machine control and human control.

This leads to another challenge to populist thinking, that everything is data-driven. There is no doubt that big data provides powerful ways of indexing and searching documents. This unstructured data capability is not suitable for workflow automation, which is highly dependent upon structured and accurate data. In reality, these rule-based documents do not contain a database as they are simply pages and pages of content. This means digitalisation of knowledge does not always need to be data driven. This is an important revelation as it avoids the expensive costs, time and risks associated with developing new types of structured data from messy sources.

There are tools now available to digitalise knowledge without any co-dependency upon data. This may not be an easy notion for some as it goes against the convention of data-driven processes established over fifty years. However, the reality is documented knowledge does not contain a database.    

These tools enable the transformation of knowledge from monologue to dialogue in the form of chatbots. It is time to look at knowledge through a new lens.