AI meant for English needs to work with such nebulous inputs without human supervision, and still be able to recognize meanings, extract associations, register the various facts, and make intelligent decisions. Such AI should also be able to inter-relate various contextual aspects and habits of users for such decisions. There are multiple other technical criteria that a data science team should use to evaluate these approaches, but perhaps they are out of scope here.

One such algorithm that is being relatively successful is called Calibrated Quantum Mesh (CQM). It is an algorithm meant for natural language and can deal with its various nuances with dexterity. CQM does away with the need for annotated training data leading to a disruptive ROI from cognitive computing projects.

These techniques are called "cognitive" because they focus on the cognitive behavior of a human when he or she engages with natural language or makes decisions based on it. However, in one respect cognitive automation is very different to human capital – it provides much greater scale, running round the clock, at a fraction of the cost of a human resource, without any manual or fatigue-related errors, cognitive bias, or risk.

Three tenets of Calibrated Quantum Mesh:

• In nature, variables do not have absolute values, but they can take various forms (quantum states) at various odds.
• Almost everything is inter-connected to everything else with various levels of influence to form an intricate mesh. 
• Patient consideration of all the available information, applied one by one, can calibrate the various odds of the various states such a mesh can take, leading to viable solutions.

The Incredible Potential of Cognitive Automation

Calibrated Quantum Mesh or any other AI algorithms meant for natural language can unlock significant value for enterprises. Particularly such algorithms can automate complex workflows that service providers deal with routinely. In absence of such cognitive automation they must rely on expensive human solution.

An example of such automation is natural language search (NLS). NLS works as if someone read through all the documents; understood all there was in images, tables or databases; figured out the intent behind a user’s question and then answered it with the right snippet – not a document, but the relevant snippet. Service providers use this technique to automate customer facing tasks like support, onboarding or other frequent processes. For example, a European service provider to automotive sector uses this capability to bubble the right knowledge to their customer service agents. It is their response to high attrition in their ranks.

Or consider an industrial major that has wrapped one of its most complicated Standard Operating Procedures within a chatbot. The chatbot educates users about the SOP step by step. It handles all the notifications and approvals from a single interface. In case of process deviations, it can gently nudge the user back to best practices. Repeated use of this chatbot also reveals bottlenecks in the design of the process, so that they can be corrected.

Another use case of cognitive automation is synthesis of actionable insights from a huge amount of free-flowing data. For instance, a large information service provider to an oil and gas major used cognitive automation for a unique use case – creating briefing packets for executives. The solution integrates with executives’ calendars, automatically detects the topics of the meetings that are coming up, and then automatically assembles everything important for that meeting from web and internal sources. The briefing pack is delivered directly to the executive, so that they are prepared.

Similarly, extraction of structured knowledge from free-flowing data and classification of information into various buckets (for data security, GDPR, etc.) are popular use cases.  In the age of Internet of Things (IoT) service providers are also using cognitive automation to merge data coming in from various sensors. Very frequently such data is not properly labeled and causes a lot of headache for downstream analysts.

The Irony of Cognitive Innovation

Innovation has always been disruptive to incumbent world order. So many powerful companies have been consumed by upstart technologies in history of business. This time, however, cognitive computing is proving to be the strongest ally of the old guard’s competitive strength against new players. 

Using cognitive computing incumbents can create experiences that are more automated and yet more personalized and more human for their customers. Algorithms can scale infinitely, which also means they can handle a multitude of parameters in making decisions. It follows that each user’s journey can be completely personalized.

The key to everything is a willingness to hop on the proverbial band-wagon. Adoption of AI and cognitive automation will be the key driver to separate successful enterprises from average ones in the next decade.