McDonald’s, the global fast-food giant, has implemented AI technology in over 500 of its locations across the United States since the beginning of 2023. Through the initiative, operational efficiencies will be gained by minimizing order errors and expediting service. A recent incident has brought that whole process into question and put a spotlight on the technology’s effectiveness, especially at peak hours.
In a strange turn of events, one customer did try to order 18,000 cups of water. Needless to say, this huge request totally rocked the AI system! As this recent incident illustrates, it can lead to unforeseen catastrophes when artificial intelligence is significantly affected by random customer behavior. One video posted on Instagram that went viral has received over 21.5 million views just on that platform alone. In the video that started it all, a customer attempted to order a large Mountain Dew. The AI would continue probing him, “And what will you drink with that?” The situation quickly turned volatile when the customer, now fed up with the bot’s continued questioning, began yelling at the AI.
While we know the intent behind this AI technology may have been noble, countless automotive dealerships have come forward with stories of trouble processing orders. As a result, customers have been subjected to some truly strange swaps. One of them found bacon in their ice cream order, and another one got hundreds of dollars’ worth of chicken nuggets delivered to them by accident. These mistakes have raised concerns about AI’s effectiveness in high-stakes environments.
Dane Mathews, a senior director for McDonald’s, admitted this bi-polaristic performance of AI. As he put it, “Sometimes I’m disappointed by it, but oftentimes it exceeds my expectations.” Her sentiment is a good indicator of the customer experience run aground when faced with an AI-centric process at quick service drive-throughs.
McDonald’s, of course, pulled its AI from its own drive-throughs last year after realizing it couldn’t reliably read and process a variety of food and drink orders. Mathews wrote about how, even though their human workers were great at handling customer orders, this is increasingly the case at peak times when time-sensitive decision-making is paramount.
Looking ahead, McDonald’s wants to continue fine-tuning the way it allows voice AI to become part of its front-line operations. Mathews underscored that the training will help teams learn how to use voice AI smartly. It will equally show when human intervention should come into play.