Henry Was Right

OpenAI selling the world as a mumblecore adaptation of Philip K Dick

On Monday, I offhandedly joked that all of the ads during football were for AI or schizophrenia medications. If you never watch live broadcasts, you may have thought this a literary flourish referencing our increasingly Philip K Dickian timeline. But no! Despite an announced MAHA crackdown on pharmaceutical advertising, Bristol Myers Squibb has been flooding the market with their creepy ad for their new schizophrenia drug. And with the pocket change they have left over from buying stakes in datacenter and chip manufacturing companies, OpenAI is running a weird new ad campaign.

I know I’m not the target audience, but I’m still flummoxed by this OpenAI ad campaign. They show twentysomethings dealing with “real life” problems like cooking or fitness. In one ad, some guy asks ChatGPT for a recipe for his date, and it opens with his girlfriend being blown away by his cooking skills.

In another, some other guy is trying to get better at pullups and asks ChatGPT for a fitness program. The struggle is real.

The ads lean into recursive nostalgia, featuring a 2010s chillwave aesthetic. They use Perfume Genius’ washed out “Fool” and Simple Minds’ 1982 classic “Someone, Somewhere (In Summertime).” I’m pretty sure they’re not targeting GenZs who all used ChatGPT to get through college. Are they going after millennials who loved Stranger Things and Washed Out? Are they chasing the aging GenX demographic that actually watched Simple Minds on MTV?

Maybe they are chasing GenXers. That would explain why these ads for AI have the same pitch as the 90s startup darling Ask Jeeves. “Good news, everyone: we’ve got websearch, but with natural language.” What makes this genre of AI ad so infuriating is they are not demonstrating anything remotely imaginative. This is not a vision of tomorrow. If anything, they are selling functionality that takes us a step backward. If you had googled “date night recipe” in 2005, you’d have found better recipes. In fact, that’s still true today:

Moreover, here’s a protip for you all. If you want fitness advice, just add “youtube” or “reddit” to the movement you’re working on:

What makes the ad campaign particularly insulting is the terrible chat responses. The date night recipe is some rehashed Sandra Lee semi-homemade pasta slop. The pull-up program doesn’t tell you how to time and progress the workouts to actually see improvement. The New York Times Cooking recipe and Reddit pull-up FAQ are far better than what the chat pukes up. OpenAI is promoting a walled garden of ungrounded, below-average search results.

The most interesting thing about this ad campaign is the response from AI superfans. If you go to the YouTube comments or Reddit, half of the people are convinced the ads are made by OpenAI’s latest Sora model. The slow panning and grainy, soft focus with shallow depth of field are somewhat reminiscent of the slop that comes out of text-to-image and text-to-video generators. Wouldn’t it be a fun trick if OpenAI produced a middling ad campaign with a couple of short prompts? Maybe that’s their secret genius here. Yes, the ads produced look amateurish and unimaginative, but the ads were made by AI! This is the worst AI advertising will ever be!

Except that’s not true. The ads were produced by people. OpenAI worked with ad agency Isle of Any, and the commercials were shot on 35mm. It’s crazy that we are in a world where someone could even think that AI could have made this commercial. But OpenAI was not intentionally leaning into the dystopian, dissociative reality they are diligently working to create. They had a different message in mind.

The production company says they were explicitly aping the Stranger Things nostalgia vibe, imagining these ads as the closing scenes of some movie. The chat response rolls out epically like the credits. “We wanted these small moments to feel elevated, and give scale to how Chat helps co-create these small moments in our lives.” I guess they like mumblecore.

So I remain puzzled. The ad campaign is intentional and deliberate, but who does this convince? They are selling the same message Google Gemini has been selling on air for a year now. It’s yet another reminder to me that outside the insanity of the Bay Area and the college classroom, this AI stuff hasn’t yet become the world-transforming, job-destroying, singularity-bringing technology that we’re constantly warned about. Maybe things will be exponentially different 14 months from now. But it’s interesting to me that despite the supergenius that is apparently needed to build these tools, none of these supergeniuses can tell a compelling story to everyone else beyond, “It’s Ask Jeeves, but 30 years later. It works this time.”

The dissonance between the broadcast vision and insular Bay Area storytelling is jarring. I know many of those enriching themselves working on these products are still convinced that white-collar jobs (like advertising) will soon be obsolete. But then why does it remain so hard to advertise interesting use cases of ChatGPT?

ETA: I couldn’t figure out how to elegantly work these citations into the main text, so I went with an oblique reference in the title, but Henry refers to Henry Farrell who presciently observed the Philip K Dick in our timeline back in 2018. I also recommend his more recent reflection on our current dystopian environment.