AI Learning must be Ongoing to Deliver Best Results
by; Russ Loignon, SVP Market Strategy and Sales
Customer meetings are a ubiquitous, fascinating part of the sales process. Always intrigued by the questions customers present when trying to resolve their unique challenges. Recently, I attended a customer sales meeting to discuss the business challenges they were facing and to determine if AI might be able to assist. During the meeting, amongst the list of potential use cases for AI, a situation with their parking garage became a leading issue. While the garage is generally reserved for customers, company employees were choosing to use the garage given the proximity of the garage to the building. Despite the garage parking fee, employees found this as an easier resolution while at the same time unwittingly causing customer issues since the garage can become filled to capacity leading to customer dissatisfaction.
The firm considered a parking service to scan employee vehicles and either remove them or ticket them. While this would alleviate customer dissatisfaction, it could create employee discontent. By using the garage video cameras and incorporating an AI solution to learn parking behaviors and patterns, creative practices could be adduced to address increased parking while maintaining both customer and employee satisfaction.
Yet, what if the patterns suddenly changed? Initially, it is likely the AI technology would enable reaction based on the learned behavior. But if the new pattern was due to external circumstances and required new or additional data to be considered, how and what would be needed to best learn while delivering new outcomes? What if the garage was suddenly less crowded? Would the AI technology assume it was managing the garage challenge with ultimate efficiency while the true root of the problem was unrelated? How would the AI technology know that the traffic pattern a mile away, was the root of the reduction in garage capacity and not AI efficiency? Surely, additional and new data would be needed to best understand this circumstance. The same could be true for human problem solving. Many times, humans misunderstand or miscalculate cause and effect. It is only through experience and learning, does human intellect determine the real situation.
In the case of traffic pattern data changes, it becomes crucial to have access to either public or private data that would identify such insight. The combination of new data, inference and explainability of all of the data are we able to absorb and interpret potential positive outcomes. Machine learning and deep learning must follow the same way. This simplistic example illustrates how AI must include as much data as possible in order to deliver the best results.
So, the next time you are stuck in traffic…. you might ponder the potential limitless ripple effect it can cause beyond the obvious.