The fast-food industry has a new AI management case. Burger King has deployed Patty, a voice assistant powered by OpenAI, in employee headsets. Staff can ask Patty operational questions—how many bacon slices on the Maple Bourbon BBQ King? How to clean the milkshake machine? When equipment fails or ingredients run out, the system updates all channels within 15 minutes without human intervention.
On the surface, Patty is an efficiency tool. In an industry with high turnover and short training cycles, it helps new employees access standards instantly and automatically manages stock alerts.
However, Patty has another function: listening to employee-customer interactions. Using keywords like “Welcome to Burger King,” “please,” and “thank you,” the AI scores each store’s friendliness, allowing managers to check performance in real-time. Essentially, your smile and tone are now judged by an algorithm.
This is not unprecedented. Amazon’s warehouse ADAPT system tracks workers’ scanning intervals precisely; falling short triggers automated warnings or even termination. Regulators in France fined Amazon for “excessive surveillance.” Call centers have deployed AI sentiment analysis, yet agents quickly learn to perform empathy cues for the algorithm rather than genuinely improving service.
In fast food, middle management is scarce and under pressure, making AI an attractive shortcut for oversight. But algorithms measure words and actions—not genuine service. Goodhart’s Law applies: once a metric becomes a target, it ceases to reflect true quality.
Burger King emphasizes that Patty is currently “assistive”, providing data for managers to interpret. Yet in the absence of capable management, assistive systems tend to drift toward “substitutive” use: scores dictate performance reviews or even employee discipline.
The lesson is clear: AI is not a substitute for effective management. Its strength lies in making management appear proactive, not in genuinely improving service. Meanwhile, employees in headsets are performing friendliness for an algorithm, not out of authentic courtesy.
