Nearly three years since McDonald’s announced it was working with IBM to develop an AI-powered order-taking system, videos have surfaced on social media showing confused and frustrated customers trying to fix comically inaccurate meals.
“Stop it! Stop it! Stop it!” two friends screamed in humorous agony in a TikTok video as an AI guide gets their order wrong, collecting 240, 250 and then 260 Chicken McNuggets.
In other videos, the AI calls a customer for nine iced teas instead of one, doesn’t explain why a customer couldn’t order Mountain Dew, and thought someone else wanted to add bacon to their ice cream.
So when McDonald’s announced in an internal email June 13, obtained by trade publication Restaurant Business, that it was ending its partnership with IBM and shutting down its AI tests at more than 100 U.S. drive-throughs, customers who had interacted with the service was probably not shocked.
IBM’s decision to abandon the deal comes as many other businesses, including its rivals, are investing in artificial intelligence, but it is an example of some of the challenges companies face as they seek to unlock the potential of the revolutionary technology.
Other fast food companies have had success with AI ordering. Last year, Wendy’s formed a partnership with Google Cloud to build its AI system. Carl’s Jr. and Taco John’s hired Presto, a voice AI company for restaurants. Panda Express has about 30 automated orders at its windows through a partnership with voice artificial intelligence company SoundHound AI.
Another SoundHound partner, White Castle, has AI assistants taking orders in 15 transports and plans to roll out 100 more, spokespeople for the two companies said. The technology completes nearly 90 percent of orders without human involvement, works efficiently with staff and reduces wait times for customers during peak hours, said Jamie Richardson, vice president at White Castle.
“It’s great for customers. it’s just as great for the team members,” he told the New York Times. “I can’t imagine why others wouldn’t invest in similar technology, but we’re very happy with ours.”
Keyvan Mohajer, SoundHound’s CEO and co-founder, believes the McDonald’s exit is just one example of a failed partnership.
“It was very clear that they’re leaving IBM, they’re not leaving voice AI,” he said. “Very quickly they go after other sellers.”
McDonald’s confirmed its intention to eventually return to this technology, writing in an internal email that “a voice ordering solution” would be in the chain’s future.
In a statement, IBM looks forward to continuing to work with McDonald’s, adding that it is “in discussions and pilots” with several restaurants interested in developing their automated ordering technology. McDonald’s confirmed the end of AI guidelines to The Times, but neither company would respond to more specific questions.
Several researchers and industry experts see McDonald’s exit as an example of how new technology has yet to live up to expectations. They doubted the company would quickly return to testing AI orders at its drive-through.
“AI systems often have this very large upfront cost,” said Neil Thompson, director of FutureTech, a research program at MIT’s Computer Science and Artificial Intelligence Laboratory. (FutureTech has worked with IBM, but Mr. Thompson said he had no inside knowledge of the McDonald’s deal.)
Currently, voice AI is often imprecise enough to require some level of human supervision, which reduces cost savings, Mr. Thompson said. And McDonald’s has a strong, higher-margin alternative offering: its mobile app.
“The app saves 100 percent of the labor involved in taking that order in a way that these AI systems, at least currently, can’t do for them,” Mr. Thompson said. “This makes it much more economically attractive for them to use the app than to use AI”
McDonald’s hasn’t abandoned all of its AI investments. In December 2023, the company announced that it was partnering with Google Cloud. A spokesman for the tech giant said it would be applied to “business use cases”, declining to be more specific.
Alex Imas, a professor of behavioral science and economics at the University of Chicago, predicted that McDonald’s will watch from the sidelines as its competitors explore the technology.
McDonald’s business model is not based on the cost savings of a few workers, Mr. Imas said. “I think they’re going to want to wait and make sure this thing is ready for commercial use.”
He expects McDonald’s to use AI in other ways, perhaps following the example of Target, which recently announced it is using the technology to help its employees.
Gee Lefevre, Presto’s interim CEO, acknowledged that the technology is very new — “less than 0.5 percent of all transits in the U.S.” are testing the use of AI to take voice commands, he said.
But he also noted that many early attempts were successful.
Wendy’s, in an email to The Times, said its artificial intelligence kiosks operate without human assistance in 86 percent of orders. And Presto had a rate of about 90 percent with most of its customers, Mr. Lefevre said.
He believes McDonald’s struggled because it used the wrong type of AI
“IBM’s model was still based on understanding natural language,” Mr. Lefevre said, explaining that the model works like a tree. When the AI hears the customer’s order, it has a limited number of branches to follow that dictate its responses and actions.
This works very well when everything goes well, Mr Lefebvre said. But in a walk-through process, where undecided customers often change their orders, he said, chains would be better off using the type of big-language model that powers chatbots like ChatGPT.
As companies continue to test their AI technologies, expect to see more videos of people taking bacon ice cream, condiments instead of food, or enough nuggets to feed a sports team.
But ask Mr. Mohajer where voice AI is going, and he’ll tell you why SoundHound has partnered with car companies like Kia and Jeep.
Imagine this.
You’re driving home from work when the car suddenly asks, “Are you hungry?”
After a few minutes of chatting with your vehicle, you decide on a burger, fries and a shake. The car finds the nearest greased spoon, places your order for you and links the directions. In three minutes, you’re up and there’s dinner, sitting patiently in a pickup lane, waiting to arrive.
