- Flip The Tortoise
- Posts
- The Most Terrible Time
The Most Terrible Time
Maybe we’re not ready for AI-generated ads. At the same time, other AI-generated content is going unnoticed.
Years ago, an industrious Google engineer built a little internal tool they called The Bumper Machine.
Google and YouTube launched a new 6-second video ad format called bumper videos, which agencies struggled to create.
Creative teams couldn’t wrap their head around how to take their painstakingly developed 30- and 60-second commercials and reduce them to just a few seconds.
The bumper machine took a longer ad and used three different machine-learning models (what Googlers would call Google AI models in 2025) to segment and cut the video into possible versions of the original that could be used as bumper ads.
On the day the internal documentation for the tool was published, I took one of my client’s ads and ran it through The Bumper Machine.
One of the outputs was absolute crap.
But the other two videos were surprisingly good.
After sharing the bumper video options that didn’t suck with my agency contacts, I ended up on a call where the creative team bemoaned the tool's future implications.
“What about the video editor’s job?”
“What about the work the strategy team and the production team put into the work?”
“What does it mean for the future of video production?”
I quipped back with one question, “Was editing the video down to 6 seconds part of the original scope of this ad’s production budget?”
The creative team's answer was a resounding “no.”
Bumper cuts were not in scope.
We ran the new machine-learning-generated ads alongside the original 30s and 60s.
The improved brand lift results of the combined set of videos created a longstanding desire to have us share future bumper options with the client and the agency in question.
Machine learning for the win!
Fast forward nearly a decade, and McDonald’s Netherlands removed a new AI-generated Christmas ad from its YouTube channel (and other social media) after facing vocal backlash in the video’s comments. Proving the adage that “sticks and stones may break my bones, but YouTube comments might destroy me.”
The video was uploaded on December 6th and subsequently removed three days later on December 9th.
I know what you’re thinking.
And I agree.
Editing a video and generating a video are vastly different exercises.
Watch the ad for yourself.
It isn’t a cinematic marvel.
And I might have omitted the animated frowning-while-burning cookies.
Maybe I’m more sensitive to the plight of oppressed animated cookies than the average McDonald’s customer.
The Sweetshop, the production company hired by TBWA\Neboko to create the ad, released a statement attempting to justify their creative efforts (which has also been taken down).
Sweetshop CEO Melanie Bridge defended the work in the press, saying that the team spent seven weeks of sleepless nights writing prompts.
“This wasn’t an AI trick. It was a film,” she said.
The thought of a creative team spending sleepless nights writing prompts for a holiday ad is somehow seasonally appropriate, perhaps more so than the ad itself.
Melanie Bridge might need to be visited by the ghosts of machine learning ads past.
Did they have visions of AI-generated sugar plums dancing in the MacBook Pros?
It also contradicts the big-tech narrative that suggests AI saves employees hours of work each week. Not at the sweatshop called The Sweetshop.
McDonald’s is not alone in the AI-backlash camp.
Coca-Cola was slammed for its AI-generated polar bears.
Last year, F1 took it on the chin for using an AI-generated “influencer.”
But I don’t understand the backlash.
Maybe people are just not ready for AI-generated ads.
This doesn’t explain why Dollar Shave Club didn’t get the same sort of reaction to its AI-generated ad (yet).
It also doesn’t explain why producing entirely AI-generated scenes in TV shows is acceptable (so far), but putting AI scenes in ads isn’t.
And it doesn’t explain why we’re so openly accepting of AI-generated music.
Recent industry research suggests that people can’t even tell the difference between AI-generated music and the real thing anymore.
I’m sure McDonald’s doesn’t feel great about the campaign.
But if it hadn’t generated such a visceral negative response, I’m not sure I’d be writing about it today.
And I’m willing to bet that the PR and news-cycle coverage McDonald’s unintentionally generated far exceeds the AI-ads’ originally planned reach.
Machine learning for the win!
I don’t know about you, but I could really go for a McDonald’s hash brown right about now.
“Every new interaction of generative AI increases the risk of serious unintended consequences.”
– António Guterres, United Nations Secretary-General
AI that works like a teammate, not a chatbot
Most “AI tools” talk... a lot. Lindy actually does the work.
It builds AI agents that handle sales, marketing, support, and more.
Describe what you need, and Lindy builds it:
“Qualify sales leads”
“Summarize customer calls”
“Draft weekly reports”
The result: agents that do the busywork while your team focuses on growth.

