DT articles
If you look at cold statistics, retailers are losing the battle against theft. Shoplifting has increased, in many jurisdictions, by over 90% since 2019. Employee theft reports have risen between 15-20% annually. In part, this helps explain rising costs at the checkout.
In the past, solutions to theft focused on hardware or human security, with modest success. Cameras and other traditional security devices act as a deterrent, but seasoned thieves know that monitoring security systems requires manpower, and retailers (particularly smaller, independent operators) do not have the financial resources to invest in these measures effectively.
Many businesses use cameras, but who monitors them? And if someone is monitoring, who has the expertise and physical presence to make the apprehension?
It would seem at first glance, that the businessperson is fighting a Sisyphus eternal battle, destined to never win. Yet, cold, impersonal Artificial Intelligence is offering the ability to fight back and roll that mythical boulder all the way up the hill.
AI generally relies on accumulated data to make informed decisions quickly. That data comes largely from human experience.
Several reports show that using AI in retail environments is very effective at theft (and fraud) reduction. The limits, however, remain. A human still has to intervene once AI has detected a problem, and without intervention, hardened thieves will ignore this new tool. Also, the old acronym of GIGO (garbage in, garbage out) applies. If insufficient data or inaccurate data is inputted, the results will be worthless.
Currently, several startups have designed AI systems that tie in with your existing camera networks. The software examines several common movements associated with theft and sends a signal, in almost real time, to the operator, who then can approach the suspect with video evidence.
The gestures and movements used to trigger the alert, though, are limited to a few dozen sequences. In real life, there are myriad “tells” or clues that indicate probable theft.
The gestures can only be detected with properly placed (and abundant) cameras and angles.
The actions, in theory, need to be captured continuously until the subject is approached, to ensure that they have not discarded the items.
In “mom and pop” operations, there may not be an opportunity o intercede before the thief leaves the premises.
Costs for the system, although reasonable, are a barrier for some micro-businesses.
These are a few of the current issues with AI detection. However, with the incredible speed at which we are seeing AI evolve, it may be only a few years before high-quality solutions present a solid barrier to theft and a strong reaction to fraud.
No responses yet