McDonald's drive-through chatbot order taker
Released: June 2021
McDonald's AI-powered automated ordering system (AOR) uses voice recognition and natural language processing to enable customers to place orders quickly and easily from the comfort of their cars.
Based on technologies from voice recognition company Apprente and AI order taking outfit Dynamic Yield, the system was trialed at ten restaurants in Chicago in 2021.
According to McDonald's CEO Chris Kempczinski, the system is reputedly 85% accurate and able to take 80% of orders. 14,000 sites across the US are expected to use the technology, Kempczinski says.
Order taking inaccuracies
Media reports indicate the company's AOR system has underperformed, with numerous TikTok users' recordings of their AI drive-thru fails going viral. One customer said the voice bot put 9 sweet teas on her order tab, when she only ordered one.
In June 2022, Restaurant Dive reported that 'McDonald’s drive-thru voice ordering accuracy rate at 24 Illinois restaurants was in the low 80% range, which is below the 95%-plus order accuracy rate the company wants before broader adoption,' citing a BITG research report.
BIPA class-action lawsuit
In May 2021, McDonald's was accused of using voice recognition to identify customers in Chicago, USA, without their consent. The fast food company is being sued (pdf) by customer Shannon Carpenter for secretly using a voice recognition system when taking his drive-through order.
The class-action lawsuit claims McDonald's is using the technology to identify repeat customers at sites across Chicago, and that it is connecting personal voice data with car license plate numbers so that customers can be more easily recognised whichever restaurant they visit.
Under Illinois' Biometric Information Privacy Act (BIPA), the collection and use of voiceprints, fingerprints, facial scans, handprints, and palm scans requires user consent.
In December 2022, McDonald's opened a new test restaurant near Fort Worth, Texas, in which all aspects of the customer experience is automated, from request to conveyer belt delivery. Only the kitchen contains human beings.
McDonald's said in a blog post that the restaurant demonstrates its commitment to 'finding new ways to serve [customers] faster and easier than ever before.' Some customers seemed to appreciate the concept, but others complained about the loss of jobs, poor pay, and lack of personal service.
Operator: McDonald's/McD Tech Labs
Developer: McDonald's/McD Tech Labs
Purpose: Personalise orders
Technology: Voice recognition; NLP/text analysis; Automated license plate/number recognition (ALPR/ANPR)
Issue: Accuracy/reliability; Privacy; Ethics; Employment - jobs
Transparency: Governance; Privacy
McDonald's (2022). 5 Things to Know About McDonald’s New Test Restaurant in Texas
Carpenter vs McDonald's (pdf)
News, commentary, analysis
Published: December 2021
Last updated: May 2023