Why McDonald’s Failed AI Automated Order Taking Project Isn’t an Example of Generative AI Failure

I keep seeing / hearing this narrative about the decision to discontinue the McDonald’s Automated Order Taking (AOT) system project, so I feel the need to jump in. The answer to that question is really, really simple. There is no generative AI in the McDonald’s AOT system. Logistically, even if it made sense that there be some use of generative AI, the solution was designed well before OpenAI released ChatGPT, or in other words, generative AI was not available when the system was designed. The recently announced Yum! Brands Taco Bell AOT initiative is also generative AI free, in case you wondered. 

There are some interesting lessons to be learned from both the McDonald’s and Taco Bell projects though. Before diving in though, let's set some context. If you think about the problem they’re trying to solve, you’ll see it is in fact, a very difficult one for several reasons. First is the environment. The environment is inherently difficult, noisy, lot’s of background noise, and a multitude of other potential issuess. The orders themselves can be complex and the way the items are ordered have an infinite number of permutations. Lastly, the customers have varying regional dialects and accents, slang phrases, and interpretations of the menu items. You’d be hard pressed to design a more difficult test for the technology.  

McDonald’s AOT Project

McDonald's decided to discontinue its trial of the AOT system developed in partnership with IBM due to multiple issues. The AI system, which had been tested in over 100 drive-thru locations, frequently encountered problems with order accuracy. Customers reported numerous errors, such as receiving incorrect items or unexpected additions to their orders. These issues were widely shared on social media, contributing to a negative perception of the technology.

Despite these challenges, McDonald's remains optimistic about the future of AI in its operations. The company acknowledged that the trial provided valuable insights and confirmed their intention to continue exploring voice ordering solutions more broadly. IBM also expressed its commitment to working with McDonald's on other projects, highlighting the potential of AI technologies to enhance restaurant operations in the future. The decision to end the AOT trial does not signify a complete abandonment of AI but rather a pause to reassess and refine their approach before potentially implementing a more reliable solution​. 

IBM utilized advanced natural language processing (NLP) technology and AI-powered customer care solutions to build McDonald's AOT system. Specifically, IBM integrated its Watson Discovery enterprise AI service, enhanced with additional NLP features, into the AOT system. This allowed the system to handle complex voice orders by understanding and processing customer requests in real time. The technology was designed to improve order accuracy and speed up service at drive-thrus.

IBM acquired McD Tech Labs from McDonald's, which had originally been created following McDonald's acquisition of the AI voice recognition startup Apprente in 2019. This acquisition aimed to leverage IBM's expertise in AI and NLP to further develop and scale the AOT technology across McDonald's locations globally. The partnership intended to incorporate multiple languages, dialects, and menu variations to enhance the customer and crew experience while maintaining high standards of security and ethical AI deployment.

Taco Bell

Yum! Brands is using advanced Voice AI technology to implement AOT systems in Taco Bell drive-thrus. This technology integrates with Taco Bell’s existing drive-thru ecosystem, which includes digital menu boards and Yum! Brands’ proprietary Poseidon point-of-sale (POS) system. The Voice AI leverages natural language processing to improve order accuracy, reduce wait times, and enhance the overall customer experience.

The technology has been fine-tuned and tested over two years in collaboration with Taco Bell franchisees, aiming to optimize operations and provide a consistent and friendly ordering experience. Additionally, this AI system is designed to ease the workload on team members, allowing them to focus more on front-of-house hospitality. Yum! Brands plans to expand this technology to hundreds of Taco Bell locations in the U.S. by the end of 2024 and potentially implement it across other brands globally in the future​

The success of Taco Bell's AOT system compared to the challenges faced by McDonald's can be attributed to several key factors:

• Extended Testing and Collaboration:

• Taco Bell has spent over two years fine-tuning and testing its Voice AI technology, incorporating feedback from franchisees to ensure that the system benefits both team members and customers​​. This extensive testing period has allowed Taco Bell to address issues and make necessary adjustments before a broader rollout.

• Integration with Existing Systems:

• Taco Bell's AOT technology is integrated with its proprietary Poseidon POS system and digital menu boards, creating a seamless ordering experience. This integration likely contributes to higher accuracy and efficiency in the ordering process​.

• Focus on Customer and Team Member Experience:

• The AI system at Taco Bell is designed to ease the workload of team members, allowing them to focus more on hospitality, while also improving order accuracy and reducing wait times for customers​​. This dual focus on enhancing both the employee and customer experience has likely contributed to the system's success. The system also does not remove the employee from the experience. Instead it takes a human oversight approach, keeping an employee engaged to ensure any issues are handled in a timely fashion.

• Holistic Approach to Technology Implementation:

• Taco Bell uses a holistic approach that leverages feedback, data, and insights to continuously improve the technology. This user-centric approach ensures that the AI system remains intuitive and user-friendly, leading to better customer satisfaction​.

• Proven Technology and Vendor Partnership:

• Taco Bell’s AOT system has benefited from leveraging existing, proven AI technologies and the expertise of its partners. In contrast, McDonald’s faced significant challenges with accuracy and user experience during its AOT trial, which led to negative customer feedback and viral social media posts highlighting the system's flaws​.

• Commitment to Innovation and Adaptation:

• Taco Bell’s commitment to continuous innovation and adaptation of its technology ensures that the system evolves based on real-world feedback and operational needs. This proactive approach to innovation likely helps in maintaining the system's relevance and effectiveness over time.

In contrast, McDonald's faced significant challenges with its AOT system, including accuracy issues and negative customer feedback, which led to the discontinuation of its partnership with IBM for this specific project​. Taco Bell's methodical approach, robust integration with existing systems, and focus on both customer and employee experiences have played pivotal roles in the successful deployment of its AOT system.