The Drive-Thru Bot Experiment: Why Fast Food’s AI Push is Stalling
Table of Contents
The promise of the frictionless order
In 2021, McDonald’s signaled a shift in the fast-food landscape by deploying AI chatbots at ten Chicago locations. The move, fueled by the 2019 acquisition of voice-tech startup Apprente and a subsequent partnership with IBM, was designed to streamline the drive-thru—the high-pressure artery of the quick-service restaurant (QSR) business. The goal was simple: faster throughput, fewer human errors, and a leaner labor model.
The trend quickly cascaded across the industry. Checkers and Rally’s rolled out AI systems via Presto across all corporate US stores in 2022. Wendy’s launched “FreshAI” in Columbus, Ohio, utilizing Google’s LLMs to understand brand-specific shorthand—knowing, for instance, that a “JBC” is a junior bacon cheeseburger. According to Wendy’s, the system achieved an 86 percent accuracy rate without human intervention during its initial phase. Taco Bell followed with similar ambitions, aiming to scale its Voice AI to hundreds of locations by the end of 2024.
The human friction point
Despite the technical benchmarks, the transition from lab to lane has been rocky. The disconnect lies in the gap between a successful “test’ and the unpredictability of a real-world customer. A January 2025 YouGov survey highlighted this tension, revealing that 55 percent of Americans still prefer a human operator over a bot, while only 4 percent actively prefer the AI interface.
This lack of consumer enthusiasm is manifesting in some unconventional ways. At Taco Bell, Chief Digital Officer Dane Mathews recently noted that the company is reevaluating its rollout after customers took to social media to troll the systems. In one notable instance, users attempted to crash the logic of the bot by ordering 18,000 water cups. Others have reported speaking in different languages or using intentionally convoluted phrasing to force the system to hand the call over to a human worker.
The fallout has already led to strategic retreats. McDonald’s ended its high-profile partnership with IBM in 2024, suggesting that the promised efficiency gains weren’t outweighing the operational friction.
The ‘Mechanical Turk’ problem
Beyond customer annoyance, the industry is facing a crisis of credibility. Presto, which powers systems for Dairy Queen, Hardee’s, and Carl’s Jr., found itself in the crosshairs of the Securities and Exchange Commission. Filings revealed a classic case of “AI theater”: much of the “automated” ordering was actually being handled by human workers in the Philippines, who stepped in behind the scenes to correct the bot’s failures in real-time.
Pivot to the back-of-house
Recognizing that the customer-facing interface is a volatile place for AI, several chains are shifting their focus toward internal logistics and “invisible” AI. McDonald’s is reportedly pivoting toward predictive maintenance—using AI to determine when an ice cream machine is likely to fail before it actually does—and implementing AI-powered scales to ensure bags are packed correctly.
Burger King is taking a different approach with “Patty,” an AI assistant integrated into employee headsets. Rather than replacing the human at the window, Patty serves as a real-time knowledge base for staff, providing recipe specs for complex items like the Texas Double Whopper. However, the system also doubles as a surveillance tool, monitoring employees for “friendliness” by tracking the use of keywords like “please” and “thank you.”
Taco Bell is also exploring dynamic menu boards that use AI to change visuals and layout on a car-by-car basis. While the specifics remain vague, the goal is to move away from a static menu and toward a personalized, data-driven sales pitch based on the customer pulling up to the sensor.
