OpenAI’s biggest advertising push yet depicts users leveraging ChatGPT in highly produced, intended-to-be-relatable everyday moments.
The only problem? Consumer research shows the effort doesn’t correlate to impact at all.
An ad titled “Dish” shows a young man attempting to impress his date with a home cooked meal by way of a cutesy prompt — “I need a recipe that says ‘I like you, but want to play it cool.’” Another, dubbed “Pull Up,” shows another young man struggling through a pull up bar exercise after prompting the bot to help him feel stronger.
Debuted during NFL Primetime, the ads are slated to run in the US and UK on traditional media — TV, streaming platforms, paid social, outdoor and influencer partnerships through the end of 2025.
But underscoring the AI industry’s formidable struggles with public perception, Adweek reports that the marketing research company System1 tested both the above ads with a panel of US consumers — and the results were absolutely dismal.
System1 measured panelists’ emotional response to the ads based on famed University of California psychologist Paul Ekman’s framework of universal emotions: happiness, surprise, sadness, fear, anger, disgust, and contempt, plus a neutral option. They also leveraged a proprietary tool to capture second-by-second emotional responses.
Although the tech industry chronically underperforms in advertising, according to Adweek‘s reporting, both ads ranked in the lowest fifth both for long-term growth and immediate sales impact. Panelists largely couldn’t even tell what brand was being advertised to them until the ChatGPT logo appears in the final seconds of the ad.
In other words, OpenAI is having a hard time connecting with real-life humans — the very consumers it desperately needs to become paying customers.
Ironically, the campaign isn’t only failing on humans. Adweek‘s Mark Ritson — who, for what it’s worth, has worked with System1 in the past — asked ChatGPT to rate one of the ads, and even the bot gave it an abject five out of ten score.
“Pull-Up is strategically on-brief and nicely made, but it underweights distinctive assets and mid-ad branding,” the bot said, “so it risks becoming a likeable, generic ‘AI-helped me’ story rather than a memorable ChatGPT ad that builds future sales.”
One thing the ads succeed at: they’re pretty to look at — but that makes sense, since they feature real people, not AI slop.
More on OpenAI: Jake Paul Invites Users to Fake Him on Sora, So They Immediately Use It to Make Him Gay and Obsessed With Makeup