Multiple KFC employees have reported that the company's newly implemented EPQC AI system frequently misaligns stock estimates with actual demand, forcing staff to navigate between system assessments and real meal preparation. What was initially expected to ease the burden on employees through digitization has instead become an additional load for frontline workers.
According to reports, previously, the overall coordination of store positions and kitchen operations relied on manual communication to confirm product needs. However, after the introduction of the EPQC system, the kitchen must follow the task assignments displayed on the screen. The issue is that the tasks provided by the system do not always align with on-site needs. For example, some employees have noticed that the system issues numerous stock tasks during low customer traffic, while during peak hours, there is a disconnect between planned and actual demand, which not only fails to improve efficiency but often disrupts the workflow.
Significant Flaws in the Operation of the EPQC AI System
In specific cases, some employees have reported that the system once requested the preparation of dozens of chicken products just before closing time. Others mentioned that the restaurant closes at 11 PM, yet at 9:30 PM, the system still required roasting 17 whole chickens. Orders during this time frame are nearly impossible to sell, and the products made are likely to end up in the trash. This series of operations has led employees to state, "It feels like we are serving the AI system."
During peak dining hours, the burden on employees is particularly pronounced. Some employees revealed to the "BUG" column that they must manage orders while also adjusting system inventory and monitoring task execution rates, which increases their workload and leads to a situation where "the busier we get, the more chaotic it becomes." To avoid disrupting normal operations, some stores have reverted to manual communication methods, with employees stating, "We still rely on people to shout, and adjust the system inventory when we have time."
The issues exposed by this system are not complex: the algorithms rely on historical data and predictive models, but customer traffic at stores is influenced by various unpredictable factors such as temporary events, weather, and nearby activities. Once there is a forecasting error, the pressure shifts to the frontline employees who must execute the tasks. More troubling is that performance metrics remain tied to the system, requiring employees to ensure that data appears correct while dealing with real-world meal preparation, leading to a "putting on a show" workaround. For the chain restaurant industry, true digitization may require the system to first learn to acknowledge the possibility of its own calculation errors.

