On This Year’s Super Bowl Ads
Charles Pulliam-Moore, writing for The Verge:
It feels like everyone who produced ad spots for this year’s Super Bowl with generative AI failed in terms of making gen AI seem useful or like something worth getting excited about. Though we’ve seen plenty of AI-generated commercials before (at previous Super Bowls, no less), this year’s event was oversaturated with them. That’s in part because image and video generation models have become more sophisticated in the past year — though still subpar compared to what humans create, they’re just improved enough for a number of brands to now be comfortable having their names associated with AI-derived footage.
Also, it’s much, much cheaper and faster to use gen AI, which is convenient when the cost for 30-second ad spots at this year’s Super Bowl ranged anywhere from $8–$10 million. With traditionally produced ads from previous Super Bowls, you could really see how spending money on production ultimately led to commercials that felt more premium than what you would usually see on television. But this year, there was an undeniable cheap and sloppy quality to many of the advertisements.
The general consensus around the advertisements during the Super Bowl has generally been that they’re distasteful. There was a time when these ads used to be amusing — the very pinnacle of American consumerism and corporate culture, almost a symbol for the Super Bowl itself. But this year, the ads generally fell into one of three categories:
- Ads for artificial intelligence companies, like Gemini, Codex, and Claude.
- Ads for sports betting and gambling products.
- Ads made by AI, like Coca-Cola and Dunkin’, to name the worst offenders.
The first category was met mostly with confusion by the masses. The Codex ad showed a first-person point-of-view shot of a student studying and doing their work before cutting to them writing prompts (seemingly on a Windows laptop) in the Codex app, which is only available for macOS. Other than that, I thought the ad was pretty good, but it seems like most people just didn’t understand the concept. And I don’t blame them — the ad says nothing about what Codex is or what you can do with it. It just uses the famous-to-Silicon-Valley-nerds slogan, “You Can Just Build Things.” Nobody who doesn’t spend time on the San Francisco side of X will ever understand that. Build what?
Claude’s ad, which I lauded before the Super Bowl, was also met with confusion, probably because it hasn’t broken out into the mainstream that ChatGPT will now show ads to free users. And even if they did know about that, hardly anyone outside of San Francisco knows what Claude is, let alone that it’s a ChatGPT competitor. The entire concept of the ad, while amusing to people who do use Claude and know the news about ChatGPT, is unintelligible to the vast majority of people watching the Super Bowl. I was incorrect in my assertion that the ad will do “surprisingly well” — if anything, my guess is proof that I, like many readers of this blog, live relatively isolated from average AI users. (Also, it came as a surprise to me that many people don’t even know ChatGPT has a paid tier.) I believe Anthropic’s ad would’ve done a lot better had it described the company’s new pledge to pay its energy costs in the places where it operates data centers.
Gemini’s ad was the best of the three, alongside Copilot’s, but those two ads were met with ridicule by most people because AI is generally unpopular among Americans, not withstanding that most office workers use it in some capacity. That especially rings true for Microsoft Copilot, which not only feels to office workers as the most likely to take their jobs, but also the one that annoys most people daily. Windows 11 is filled to the brim with AI slop interfering with people’s computing lives, and I’m unsurprised the tide has shifted on its adoption. People are sick of Copilot. Most people, meanwhile, blissfully use the often-incorrect Google AI Overviews at the top of search results without even knowing they’re powered by Gemini. I don’t think Google’s ad clarified that at all.
I also want to touch on what I’ve been calling the grifter ads, from the sports gambling companies to “AI.com,” a company owned by Crypto.com that ran an obnoxious ad campaign telling people to go to its website and reserve a handle, before plugging “ai.com/elon” and “ai.com/sam” at the end. “Who’s Sam?” has been a frequent question among the normal crowd. The company’s purely vibe-coded website crashed immediately after the ad aired, and despite spending $70 million on the domain, it still lacks a favicon. Regardless, the app is a scheme to get users’ email addresses and credit card data; it is advertised as a way to host OpenClaw agents. I signed up for an account and used my Apple Cash card with no balance as a safety measure, and I’m yet to hear anything back. Typical crypto scam. If anything, this is just proof that unpopular con artists are raising way too much money in the Valley.
The sports betting companies are also obnoxious and signal a rampant gambling addiction in the United States. Much like the AI industry, they’re entirely unregulated casinos now marketing to the entire country dozens of times during the most-watched telecast of the year. These companies are morally indefensible, much like the gambling and insider-trading markets Polymarket and Kalshi, which were barred from advertising during the game this year. (I predict this rule will be removed next year amidst lobbying efforts from the venture capital-backed and grifter-run firms.) Repugnant ads promoting serial gambling on sports games are objectively bad for society, and it’s unsurprising but equally callous that they haven’t been kicked off the airwaves.
Back to Pulliam-Moore’s point, the AI-generated or AI-augmented ads were also in poor taste. My least-favorite was Dunkin’s, which de-aged a host of famous sitcom actors like Matt LeBlanc and Jennifer Aniston of “Friends” to promote doughnut holes or something. Seriously, what is going on here? I think it would’ve been adorable and amusing to see the real, non-AI-edited actors in a commercial, and that would’ve almost certainly done well. But business executives at top corporations — who are almost all Generation X or Baby Boomers — fail to understand that the use of AI in creative works is loathsome. The Pepsi polar bear ad did much better than Coca-Cola’s because people saw that the Pepsi one was genuine. It might’ve been computer-generated imagery, but it was drawn by a human being. The Coca-Cola ad was pure slop in the way the Pepsi one wasn’t, and people were able to discern that.
If anything, the Super Bowl taught us that corporations are woefully out of touch with everyday Americans. Coinbase’s ad was booed, nobody understood Claude and Codex’s, people don’t even know what Gemini is, “AI.com” is a grift nobody’s ever heard of, and people hate AI in ads. The Super Bowl was a mess of enigmatic corporate AI slop that almost everyone was uninterested in. But if Silicon Valley is to be believed, by next year we’ll have AI agents watching and summarizing the ads for us. What a ridiculous industry.