Anyone who works in the retail industry is aware of a term known as “shrinkage”. Shrinkage is normally defined as a loss of inventory for other reasons than poor sales alone; theft being a common example. Might AI-powered systems be able to reduce shrinkage? Let’s take a closer look.
The Problem at Hand
According to recent statistics, shrinkage costs the United States retail sector alone more than $94 billion dollars every year. While this is concerning, other factors come into play and we are not only referring to customer theft. Additional reasons may include poor inventory management, shipping errors, and even vendor fraud.
Before moving on to discuss how AI surveillance software may be able to provide innovative solutions, it is wise to quickly examine why older approaches to shrinkage no longer work as they should.
The Issue with Traditional Methods
There are several problems directly associated with traditional methods of surveillance and quality control. Perhaps the most prevalent involves the fact that we are only human. Employees cannot detect every fraudulent activity. Some individuals are more keen to notice suspicious behaviours when compared to others. A handful of workers also might not take the notion of shrinkage as seriously as they should.
A second confounding factor involves developing a means by which an entire facility can be monitored on a 24/7 basis. Not only is this a very complicated task, but many businesses do not possess the funds to implement such networks. The good news is that the rise of artificial intelligence (AI) is already providing a handful of unique solutions.
The Use of AI Algorithms
Some readers may already be a bit familiar with artificial intelligence. After all, our lives are already being affected by its presence. Common examples include (but are not limited to):
- Personalised online search engine results.
- Smart devices
- Chatbots
- Facial recognition software
Although the processes themselves can be quite complicated, they all tend to rely upon a series of commands generally known as algorithms.
Algorithms are essentially instructions which tell a program to behave in a certain way when presented with a specific situation. As algorithms have become extremely advanced, AI systems are now able to perform complex tasks and even mimic human behaviour to an extent. So, how do algorithms fit into surveillance technology?
Anomaly Detection
Imagine for a moment that you are a detective tasked with reviewing a videotape of a potential theft. You will normally be looking for several cues such as nervous movements, what type of clothing the individual is wearing, and where his or her hands are placed.
AI can perform these very same analyses and provide results in a matter of seconds. Furthermore, these systems are capable of simultaneously monitoring numerous individuals or even crowds. This provides stakeholders with a much-needed edge over potential thieves and shoplifters. In the event that an anomaly raises a “red flag”, the situation can be immediately brought to the attention of security personnel.
More Than On-Site Theft
As mentioned previously, shrinkage does not necessarily have to result from instances of theft. It may be accidental (such as operational loss or administrative errors). There have even been situations when the vendors themselves are to blame. AI can once again help to address these situations.
We need to remember that AI-powered security protocols go far beyond real-time cameras and CCTV systems. They are also employed within other areas of quality assurance including data entry, pipeline management, and inventory control. Note that artificial intelligence is not necessarily intended to prevent human intervention. It is instead designed to provide an additional degree of oversight.
These are some of the main reasons why AI-powered systems have already enjoyed a significant presence and yet, another question needs to be asked. Just how reliable is this software?
A Work in Progress
Interestingly enough, studies have indicated that AI is already more than 86 per cent accurate when detecting theft. It is even more useful in terms of operational oversight. Still, these systems are not perfect by any means. So-called “false positives” can occur on occasion. Furthermore, not every business can afford this technology at the present. There are also questions involving humans who rely too heavily upon AI.
Although these can be considered shortcomings, we need to remember that AI is constantly evolving. As it becomes even more advanced, there is virtually no doubt that it will provide a host of benefits to the retail sector as a whole. All that is required is a bit of patience.
