©RDNE Stock project via Canva.com
June 18, 2024
Can Fast-Food Drive-Thru AI Become Mainstream?
McDonald’s is ending its AI-driven drive-thru ordering experiment with IBM by late July 2024. Despite the end of this partnership, McDonald’s believes AI voice ordering has potential for the future and will continue exploring new solutions for customers who prefer this style of service, according to Restaurant Business.
Mason Smoot, McDonald’s USA chief restaurant officer, acknowledged some successes but emphasized the need for broader exploration of AI solutions. While reasons for ending the IBM deal are unclear, McDonald’s remains open to future AI solutions for drive-thru orders, suggesting the potential for new partnerships to enhance restaurant technology.
One potential reason for the end of the partnership could be the accuracy of AI orders. Customers reported a variety of errors at McDonald’s drive-thrus, including receiving an ice cream cone topped with bacon and being charged for nine sweet teas instead of a single hash brown, sweet tea, and Coke. Another instance involved a customer being billed $222 for multiple McNugget meals due to AI miscommunication.
These incidents were widely shared on social media, particularly TikTok, highlighting the system’s flaws and contributing to the negative feedback. One TikToker posted a video with a text overlay that says “fighting with McDonald’s robot.”
The AI-driven order-taking system, although facing termination at McDonald’s, has found traction among other fast-food chains. Companies such as White Castle, Checkers, Rally’s, Hardee’s, Carl’s Jr., Krystal, Wendy’s, Dunkin’, and Taco John’s have either tested or fully implemented similar technologies. Domino’s employs AI for voice recognition in order-taking and optimizing delivery routes, while Starbucks uses AI to personalize customer recommendations via its app. In China, KFC has adopted AI facial recognition technology to predict customer orders, showcasing the diverse applications of AI in the fast-food industry.
Furthermore, many fast-food chains are exploring artificial intelligence to boost sales and efficiency, particularly in drive-thru operations. Presto Automation provides AI voice recognition technology to several major chains, including Checkers, Rally’s, Del Taco, Carl’s Jr., and Hardee’s, as well as digital ordering tablets to casual dining restaurants. However, Presto is facing financial challenges, including a $6 million debt repayment, and has lost some key clients.
Del Taco initially praised Presto’s AI for drive-thru orders, stating that it was “exceeding expectations,” but has since decided to discontinue its use. Presto’s AI voice recognition, though advanced, still requires human intervention for order input and accuracy, managing only 30% of orders independently.
Moreover, Presto Automation faces scrutiny as filings reveal that over 70% of orders are completed by off-site human workers, notably in the Philippines. This practice contrasts with the company’s AI branding and raises concerns about transparency in AI-driven solutions across industries. Despite claims of automation, the reliance on human contractors raises questions about the true impact on labor in the restaurant industry’s adoption of AI technologies.
Meanwhile, White Castle believes its AI voice-recognition technology surpasses human accuracy. White Castle has implemented SoundHound’s AI voice system in 15 of its restaurants for drive-thru orders, aiming to enhance accuracy and efficiency while reducing staff pressure and customer rush. The technology, which White Castle plans to expand to over 100 locations by year-end, allows customers to place orders via an AI named Julia. Despite concerns about AI reliability and its impact on jobs, White Castle reports that the system has not reduced staff hours but rather allows employees to focus on food preparation and customer service.
In December, Bloomberg reported that McDonald’s was in discussions regarding a potential deal with Google, involving a chatbot named “Ask Pickles” that employees could use for various tasks. McDonald’s has also continued to integrate automation across its operations in other ways, including mobile ordering, in-store kiosks, drone deliveries, kitchen robotics, and AI-driven hiring tools.
Discussion Questions
How can fast-food chains balance AI-driven technologies like voice-ordering systems with maintaining high order accuracy and customer satisfaction, considering recent challenges?
What strategies should restaurant chains adopt to manage customer expectations and mitigate risks associated with AI failures or miscommunications in drive-thru operations?
Poll
Loading …BrainTrust
Brian Numainville
Principal, The Feedback Group
Melissa Minkow
Director, Retail Strategy, CI&T
Boran Cakir
Co-Founder & CEO, frnt
Recent Discussions
More Discussions
Will an In-Store Retail Tech ‘Confidence Gap’ Create Winners and Losers?
A new survey suggests that in-store retail tech could be a major factor in retailer success…Walmart’s Next CEO Era: From McMillon to Furner
Walmart CEO Doug McMillon will soon be replaced by John Furner. What’s next for the retailer?Which Brands Are Missing the Boat on Advent Calendars?
Advent calendars appear to be trending upward in popularity, with potential to move to other holidays…
Anyone who has a voice assistant like Alexa or Siri knows that the propensity for mishearing or misunderstanding is high. And so it was with McDonald’s AI voice ordering: too many mistakes, too many odd things being added to orders. Amusing as it might be to hear about on TikTok, those errors are a disaster in fast food where efficiency and accuracy are vital to protecting margins. All this said, I don’t think it’s the end for AI ordering and voice ordering in drive-thru: the technology just needs to get better, which it will do over time.
At present, AI is still mistake prone, but like all other technology it will keep improving and eventually perform almost perfectly. Even with McDonalds setting it aside for right now, there will be a deep and wide future in fast foods, including at McDonalds.
Thanks to the salary pressure brought about by protesters, fast food management will covet any technology that replaces employee-salaries. Of course, another benefit is that AI shows up for work on time every day.
However, much like humans, AI will make customers mad if it doesn’t get the orders just right, even when the confusion was caused by customers by mumbling, mispronouncing words, wrongly agreeing to what they didn’t really want, etc. – Db
We are just at the beginning of the use of AI technology for applications like drive-thru ordering. While it may not be ready for prime time in some of these cases, the technology will improve and be used in more and more cases in the not all that distant future.
Yes, a lot more testing needs to be done before it can be adopted into the mainstream. Ice cream with bacon is hard to imagine but once logic and learning are applied and it can become more intuitive, then maybe it can be a home run. Too much human interaction is still required.
We already know that McDonald’s fries are great when dipped in a shake, and the chili crisp with ice cream trend is hot – so soft-serve with bacon bits on top sounds like a winner.
Could be like a poutine: bed of fries, soft-serve on top of that, sprinkled with bacon. Oh, I’m getting hungry…
While brands can’t risk getting so much wrong with customers, it’s important not to give up on AI, as it gets better over time. Restaurants and brands simply have to work on improving the AI behind the scenes, before it makes its way to consumers, rather than having the experiments cause so much chaos for customers.
So that’s a problem?? 🙂 Seriously tho, since this is still in its infancy, I’m relucatant to draw many conclusions. But once it’s “perfected”, we can look forward to…what exactly? The process will still be a long way from automated, so while this might automate order taking – i.e. the front part of the process – I’m unclear on what else it will accomplish. And in low volume situations, where order taking and fulfillment are done by the same person, I don’t see much in the way of savings; which is unfortunate, because it seems like a net loss in the human interaction factor.
AI is progressing faster than ever. Newer – and more capable – technologies featuring voice recognition are better than ever. The problem may be background noise. Just like an employee reads back the order, so should the AI-fueled order-taking technology. In short, catch the error before the customer leaves the ordering system.
Today, AI is a solution desperately seeking an application. Drive thru order taking is currently one of the AI targets because they’re hoping it’s simple enough for their solution to be valuable. But note what is key there: AI isn’t anywhere close to what is being promised from it. So be careful before buying into the idea that drive thru is an ideal application. McDonald’s evidence is that quality isn’t there yet. And nothing irritates customers faster than frustrating order processes. AND, the real key question is why anyone would think this is a business opportunity – a way to drive bigger profits? I just don’t see any benefit of significance but the great risk in these systems is clear.
A non-starter from the earliest beginnings.
I have to agree. The AI hype factory is in overdrive. While I think AI will change many things, the technology isn’t always what it’s cracked up to be at this moment in time. Ordering via an app on your phone is still way more reliable than ordering via voice powered by AI!
The use of AI to optimise operations, such as drive-thrus, is still in its infancy. Naturally, there will be some issues along the way. Over time, these issues will become negligible, and you should expect faster service for customers (less waiting time) and decreased costs for fast-food chains.
While we are going through this early adoption phase, fast-food chains can balance AI-driven technologies by better using machine learning to improve AI’s recognition of accents and background noise in controlled environments. They should provide backup customer service options if there are any AI issues (confirmation if the order is correct and passed on to restaurant staff if the issue persists). Additionally, businesses have a responsibility to educate customers about AI capabilities and limitations while setting realistic expectations to mitigate potential frustrations. From initial deployments, there has been very little, if any, of this.
The import of the application of AI technology in drive-thru QSRs is only to the retailer. It’s intended to cut down on payroll expense. Unlike the use of AI in Ecommerce, which can streamline the customer experience, this use does not noticeably improve the customer service. The consumer doesn’t need it. If this use of AI went away tomorrow for good, no consumer worth his burger and fries would care.
While McDonald’s discontinue the trial, it is important to note that “McDonald’s believes AI voice ordering has potential for the future and will continue exploring new solutions for customers who prefer this style of service. Yes, there are missteps with AI voice ordering, but humans make mistakes too. I am confident that AI voice ordering accuracy will improve and become mainstream in the near future.
When it comes to AI-driven technologies at fast-food chains, don’t throw the baby out with the bath water! These early stumbles are regrettable, but all part of the process. These systems continue to learn and get better, and while messing up some orders looks bad, but it’s only a matter of time. And let’s not forget that order taking is only one application. The use of robotics for preparing food is already gaining traction. Many of the restaurant chains are doing exactly what they should be doing: testing, learning, failing, adjusting and testing some more.
I agree, but I also think the roll out of this was way too early. These kind of glitches could surely have been discovered in pre-testing before it went into field. It’s not good business practice to frustrate customers with beta version of poorly performing technology.
To balance order accuracy and customer satisfaction, fast food chains can phase in AI as a collaborative tool to help rather than supplant human workers. While AI boosts process efficiency, human verification ensures accuracy, adds warmth and protects the integrity of the guest experience.
Give me a real person who is invested in their job anytime over an AI driven interaction. Will this happen? Sure-eventually-but not for a while, simply because the technology is not ready. Just consider the ridiculous waste of time phone bots are-STILL.
The TikTok video embedded in the article provides at least part of the answer to mitigating risks associated with AI failures. Once the system became confused, an alert or override was used to address the customer directly. In addition, providing as much transparency as possible to the customer around the trial is crucial. Letting customers know they’re in the privileged position of testing out a new system, that it may not get everything right, and also welcoming feedback may reduce the negative backlash on social media. It brings the consumer along on the journey, and asks them to help.
I am guessing that every one of us has experienced order errors from humans when using drive-through.
AI ordering is here, and it is just a matter of time before its development will generate perfect results. Like with people, AI must be trained. Mistakes are the best training tool.
I didn’t know McD’s sells a bacon-topped ice cream cone. I am interested. However, should the human have stopped it unless it is on the menu?
Why can’t AI partner with a human for ordering? The human takes the initial order, the AI system confirms it with the customer, perhaps tries for an upsell, while the attendant is already onto the next customer?