McDonald's AI Drive-Thru debacle is a warning to us all

We've mentioned before the risks for brands jumping on the AI bandwagon too quickly. And that extends beyond using AI image generators to any kind of customer-facing application, as McDonald's may have learned from its AI Drive Thru fiasco.

AI technology is advancing rapidly but remains in a state of relative infancy, and in many cases it just isn't good enough yet to implement without causing significant friction. The world's biggest fastfood brand has sensibly decided not to extend the contract on an AI voice recognition service and has told franchisees to remove the tech, but did it not think it should at least test it before it became the subject of viral videos?

Developed by IBM, McDonald's AI ordering system was implemented in over 100 McDonald's locations in the US starting back 2021. It was supposed to use voice recognition to process orders, but customers reported frequent frustrations, including quite spectacular order mixups, from bacon being added to ice cream to orders being hugely inflated.

In one video shared on TikTok with the caption "Fighting with McDonald's robot", the AI interpreted a woman's request for vanilla ice cream and a bottle of water to be an order for a caramel sundae and multiple sachets of ketchup and butter. In another, a customer inadvertently ordered 2,510 McNuggets Meals. That left a human attendant to have to reinput the order, rendering the AI a pointless frustration.

As reported by the trade publication Restaurant Business , McDonald's is removing the tech but remains determined to push forward with voice recognition technology to avoid having to employ humans to do the job of taking orders. The company said in a statement: "While there have been successes to date, we feel there is an opportunity to explore voice ordering solutions more broadly.

"After a thoughtful review, McDonald's has decided to end our current partnership with IBM on AOT (automated order taking) and the technology will be shut off in all restaurants currently testing it no later than 26 July, 2024."

This is far from the first case we've seen of experiments with AI resulting in a customer backlash. Lego is one of several brands to have apologised after using AI imagery. We've also seen branding agencies warn against AI washing , which is a tendency for companies to overstate their AI capabilities in order to make themselves look like part of the zeitgeist.