The term AI has been one of the most highly referred to concepts/strategic initiatives in the evolving digital era. Despite the continuous references of AI by many across industry sectors, the true meaning of the concept may still allude even the most sophisticated executives. I’m not going to try to provide a total clarification of the topic here, but I will introduce some basic essentials that may help provide a baseline understanding in how to or whether to adopt AI in an organization.
Let’s start with a little background on the concept of Artificial Intelligence. This technology has actually been in existence for decades with such components such as neural networks having been utilized for pattern recognition in data resources for a multitude of business applications. The reason why AI has reached the current ultra-focus is simply due to the advancements in computer processing power and speed and an abundance of data resources being generated on a daily basis. These two factors have unleashed the powerful capabilities of this data centric, computational methodology. To put this in perspective just consider the “room sized” main frame computers of yester-year that were required to process data, now compare that to the capabilities of what today’s pocket sized smart phones can do.
So what does AI do? And where can it be deployed??
Replication of Tasks and Automation
Generally there are two types of scenarios for which AI can play a significant role in enhancing business processes. The first, and more highlighted situation, involves the replication of repetitive tasks and processes which are rules based in nature. What this refers to is any type of activity that can be defined as a set of sub-component tasks that can be replicated by computer code. Simple examples include:
If consumer responds “X”…then perform “Y”
If bar code reads “A”…then place product in “B” location
If responder reacts with “C”….then initiate “Z”
In other words, activities that involve a series of tasks that can be described by “if then” type of code can be replicated by AI. The power of this lies in the amount of processes that can fall into this category These types of scenarios can involve customer support & service, inventory management, query routing, transaction facilitating, compliance issues, etc. The power of today’s computer processing and AI capabilities that are faster and can process many more possible data attributes of processes can more fully define and replicate these processes and create an automated reliable system. Consider a major player Wyndham Hotels and Resorts that pursued process automation in a number of their activities.
The key to these task replicating systems is the ability to map out the sub-component interactions that comprise these tasks. In other words, recording the underlying steps and rules that comprise the full activity based structure of these tasks. This data feeds AI algorithms that learn and replicate the process. The real power of AI is when truly complex tasks are sought to be replicated, where the recording and mapping of rules entails copious possibilities.
The other prominent incorporation of AI is for predictive modeling which has also increased dramatically in the digital era. This application is a bit trickier in that the scenarios often include variances or unknowns in what defines movement or performance of designated metrics that need to be understood and forecasted (often these are KPIs). For instance, as a strategist, you may want to predict future sales of a product or consumer response to a marketing initiative or the potential of fraud. These and many more scenarios can include a multitude of driver or descriptive data variables that cause these key indicators to move in one direction or another. AI can be applied but also a host of more traditional methods as well such as regression based methods that can be used to create models to help identify and describe how data variables interact to replicate the past in order to perform “what if” simulations and predict the future.
A major differentiator between AI automated tasks and AI for prediction is the presence or lack of unknown variance of what defines a process. For automation, AI will consider all the possible interactions that comprise a process and replicate it exactly. In prediction modeling, AI and other methods seek to identify patterns that best explain what drives metrics, but there generally exists a source of unknown variance that models can’t capture. These models seek to minimize the variance of unknowns to enhance prediction.
Stephan Kudyba is founder of the analytic solutions company, Null Sigma Inc., which focuses on providing strategic analytic solutions for organizations across industry sectors. He is also a professor in the management department at New Jersey Institute of Technology where he teaches courses that address the utilization of IT, advanced quantitative methods, business intelligence, and information and knowledge management to enhance organizational efficiency. He has published seven books and numerous journal and magazine articles on strategic utilization of data, technologies and analytics to enhance organizational productivity and has held management and executive positions in leading organizations. Dr. Kudyba holds an MBA from Lehigh University and PhD in economics from Rensselaer Polytechnic Institute.