The ambitious AI drive-through experiments by major fast food chains McDonald’s (NYSE: MCD) and Taco Bell (NASDAQ: YUM) have recently hit significant roadblocks, revealing that artificial intelligence is still far from flawless in handling everyday customer interactions. After several years of trials and expansion, each chain has faced notable issues that have forced them to reconsider or even halt their AI-powered ordering systems.
McDonald’s launched its AI drive-through ordering system as a partnership with IBM in 2021, testing the technology in over 100 locations. The concept was appealing: allow an AI to take orders to reduce wait times and ease the workload on human staff. But the experience quickly veered into chaos for many customers. Viral videos surfaced showing the AI wildly misunderstanding orders, one woman’s request to remove a Diet Coke resulted in nine sweet teas being added, while another customer’s order ballooned into hundreds of dollars of chicken nuggets. These errors generated frustration and laughter online, but for McDonald’s, the mishaps underscored serious limitations in the technology’s current capabilities.
By mid-2024, McDonald’s decided to end the IBM AI drive-through order tests completely, with systems shut down by late July 2024. Mason Smoot, McDonald’s USA Chief Restaurant Officer, communicated to franchisees that despite some successes, there is “an opportunity to explore voice ordering solutions more broadly.” The company emphasized that it is not closing the book on AI voice ordering altogether. Instead, they plan to evaluate new solutions going forward, hopeful that future technology can overcome the current hurdles. IBM echoed confidence in their AI capabilities but supported McDonald’s decision to pause and reassess the project.
Meanwhile, Taco Bell aggressively expanded its AI voice ordering system starting early 2023, rolling it out in more than 500 locations across the United States. The chain intended for the AI to reduce wait times and improve order accuracy while easing strain on employees. But the system was not immune to problems. One viral incident involved a customer ordering an absurd 18,000 cups of water, effectively overloading the AI system. Other clips showed customers growing visibly annoyed as the AI repeatedly prompted them to add unwanted items. These glitches created frustration and drew negative attention on social media.
By August 2025, Taco Bell’s leadership, including Chief Digital and Technology Officer Dane Mathews, publicly admitted the technology had its ups and downs. Although there were moments when the AI worked surprisingly well, frequent errors and customer dissatisfaction prompted the company to slow its AI rollout and reassess its approach. Taco Bell now plans to guide employees on when to rely on the AI system and when human intervention is likely necessary, recognizing that some aspects of customer service still require a human touch, especially during busy periods.
In contrast, Wendy’s has taken a different path with their AI drive-through efforts, seemingly achieving more promising results. Partnering with Google Cloud and Accenture, Wendy’s launched FreshAi, a generative AI assistant designed to better handle the complexity of orders at scale. This technology helps crew members by accurately interpreting natural language, understanding numerous menu variations, and providing real-time support during busy periods. Wendy’s has expanded the use of FreshAi to hundreds of locations and reports improved order accuracy and service speed thanks to AI assistance, without the high-profile blunders seen at other chains.
These contrasting experiences highlight the challenges in deploying AI for complex, real-world tasks involving natural human speech and the unpredictable nature of customers. While the idea of AI-driven drive-through service is attractive for its potential efficiencies, current systems are clearly still learning how to deal with nuances, accents, and the occasional prankster intent on overwhelming the technology.
For now, McDonald’s has stepped back to explore more robust solutions for voice ordering, and Taco Bell is adjusting its deployment for a balanced integration of AI and human oversight. Wendy’s progress suggests potential in generative AI, but widespread flawless AI ordering remains a goal rather than a reality at this stage.
This ongoing saga serves as a reminder that AI technology, especially in consumer-facing roles requiring complex communication, still has a long way to go. The path ahead will likely include continued experimentation, human-AI collaboration, and careful calibration to deliver improvements without sacrificing the customer experience.
