I first saw this sad warehouse post on Tuesday as I was preparing for a panel discussion on AI use and ethics. X user Chris Alsikkan posted two images of him. One of them shows him a fantastic dreamscape of peppermints and a giant lollipop the size of a sofa. Inside, whimsical lights reflect off the walls and gumdrops guide guests through a magical tunnel. The other appears to be a favor from a disastrous children’s party, the remains of which were left next to the trash can along with childhood hopes and dreams.
The Willy’s Chocolate Experience event in Glasgow was billed as ‘the place where chocolate dreams become reality’. In fact, it was there that the dream turned into a nightmare, the police were called, and many news outlets reported on it, including the New York Times.
I’ve been researching and writing about the use of visual artificial intelligence for the past six months. I’ve tried it myself, creating illustrations for educational and presentation purposes. I’ve watched companies struggle to stop disinformation and read about people dealing with the horrific aftermath of nudity apps used against teens and women.
These are all really big questions to address. But I can tell you that there is no world in which I thought a silly story like Willy’s chocolate experience would help illustrate the major AI problems we face.
If you stop and think about it, that’s not surprising. I am drawn to the many stock images of events and gorgeous descriptions of the great time we are about to have. With the quirky, bright, and colorful images the site featured, it’s no wonder people lined up in hopes of seeing something resembling a Tim Burton movie set.

(Screenshot/Willy’s Chocolate Experience)
But for those of us who pay attention to AI, it’s clear that the site was embedded with AI-generated illustrations. Most image generation tools have similar default digital art styles. They tend to be very colorful and packed with visual content. Wild misspellings are a killer for Willy’s site.
But the general public still doesn’t pay as much attention to AI as we do. To be honest, the illustrations are a representation of what to expect at the event. Spelling mistake? Perhaps it’s a fanciful homage to Roald Dahl. Who wouldn’t want an “Exerdray He Lollipop”?

(Screenshot/Willy’s Chocolate Experience)
In the 1971 film version, Wonka, played by Gene Wilder, tells visitors, “There’s a little surprise around every corner, but nothing dangerous.” AI will cause the opposite, with many big surprises and big scary problems hiding around every corner. But I never thought event organizers would need to use technology to protect themselves from over-promising and under-delivering.
While listening to a recent episode of “Offline with Jon Favreau,” Favreau and international reporter and columnist Max Fischer discuss how scammers created a deepfake of Selena Gomez giving away Le Creuset cookware. We talked about what we had created. It’s very easy to see how people fall for scams. Gomez has a cooking show on Max. That means she has a lot of footage of her that can easily be used to train AI models. Combine that with her fans’ parasocial relationships and it’s a recipe for manipulation. Even that scam was just a clone of the scam using Taylor Swift.
It’s always been possible to create content like this, but it doesn’t take much time to develop a website, illustrate art, manipulate images in Photoshop (either paid or illegally copied), or take the time to copy them. I needed the skills to write, shoot and edit videos. Create and post social campaigns. These are all different skill sets that people develop over their careers. Now it can easily be done by one person in a few hours at most. All you need is a credit card and pure imagination.
I’m concerned about how quickly technology is changing and how it’s nearly impossible to keep up, report on it, and build guardrails. What I should have been prepared for, but what I didn’t expect, was how quickly people would figure out how to use it to scam people.
In the case of Willy’s Chocolate Experience, some parents had a terrible time wasting their hard-earned money. In the Gomez scam, people rack up monthly charges they don’t authorize while spending valuable time disputing credit card charges or waiting by the mailbox for a new cookware set that never arrives. .
We need to keep these examples in mind as we continue to report, analyze, and discuss AI. There are huge social impacts that deeply harm people, but there are also smaller things that we may not think about or cover as much that can be just as harmful.
It is my (slight) hope that AI is properly regulated, researched, and permanently deployed, but I fear that Willie has already gone off the ramp. My hope is nothing more than a papier-mâché rainbow held together with Scotch tape and staples.