While the battle on the gridiron captured the world’s attention, the battle for AI supremacy raged on right alongside it—in the commercials. We sat down with Neeraj Arora, Arthur C. Nielsen, Jr. Chair in Marketing Research and Education and Matt Seitz, Director of the AI Hub for Business at the Wisconsin School of Business, to break down the strategies behind the $8 million Super Bowl spots from AI heavyweights, OpenAI, Anthropic, Google, and more. A researcher and practitioner, Arora and Seitz shared the promise and pitfalls of different advertising approaches for these tech titans along with key takeaways for marketers.
The Great (Or Not So Great?) Anthropic “Attack Ads“
The most discussed moment of the night was Anthropic’s decision to take a direct swipe at OpenAI. One ad featured an overeager chatbot-turned-salesman with the cheeky tagline: “Ads are coming to AI. But not to Claude.” This is likely in response to OpenAI’s recent announcement that they would start advertising on their ChatGPT platform.
Arora found this to be effective positioning from Anthropic, a challenger brand. “They’re differentiating themselves. ‘This is who we are. We are very product-focused. We are not going to sell ads.’ He notes that Anthropic has maintained a clear strategy with rapid, exponential growth, while questions remain about OpenAI’s identity.
Seitz, however, questioned if the ad was “off-strategy,” especially if Anthropic is after corporate clients. For a company aiming for enterprise business, is a potentially, “mean-spirited” attack the right way to build trust and gain credibility?
“They’ve been trying to take the high road. They’re the ‘responsible’ company. And then they go and run an attack ad on the leader in the market at the moment we’re defining a new category.”
However, research shows that going negative tends to stick. Says Arora, “It’s pretty well documented. Negative advertising just pops. You look at our politicians, right? The reason they are generally very negative about each other is because there’s research showing that it works for good—for better or worse.”
The consensus? There isn’t one yet. Both experts acknowledged the ad’s effectiveness at generating attention and clearly differentiating Claude from ChatGPT. But whether this bold move translates to the right kind of customers (and whether it’s worth potentially undermining years of brand positioning) remains to be seen.
Tip for marketers: Attack ads may generate buzz, but they risk undermining carefully built brand positioning, particularly when a category is still being defined.
Watch Anthropic’s “Can I get a six pack quickly?”
Can I get a six pack quickly?
A Time and Place for “Slop?”
Is it AI slop—the notion of low-quality AI-generated content—or the uncanny valley that left Arora and Seitz uninspired by Xfinity’s Jurassic Park send-up and Dunkin’s “Good Will Dunkin’” spot, which both featured de-aged celebrities. Similarly, in 2024, Coca Cola took a hit to their brand equity by using AI to create the company’s annual holiday ad, featuring their iconic polar bears. Arora and Seitz explored the relationships between these ads.
Seitz draws a parallel to auto-tune in music: once hated, now largely accepted. “Do you think about auto-tuners?” he asked students recently. “They said ‘we’re fine with them.’ But they hated the AI Coca-Cola ads. So what’s the difference?”
Arora personally found the Dunkin’ ad “creepy” rather than funny, noting concerns about authenticity. He dove deeper, adding, “There’s a more nuanced possibility that might be emerging…where it’s entirely obvious that you know this is AI generated and it’s funny, so viewers are fine with it. But in other situations when you’re building a brand and you’re trying to be distinctive that it’s not going to work.”
“There’s a more nuanced possibility that might be emerging…where it’s entirely obvious that you know this is AI generated and it’s funny, so viewers are fine with it. But in other situations when you’re building a brand and you’re trying to be distinctive that it’s not going to work.”
Neeraj Arora
Interestingly, even vodka brand, SVEDKA, leaned heavily into machine and tech connotations. In their case, the ad—featuring slick, dancing cyborgs at a night club—leans heavily on tropes we’ve come to expect in alcohol ads—parties, humor, fun, with a techno twist. While the ad appeared AI-generated, the execution served their purpose well, allowing for quick brand awareness and name recognition.
The consensus? While using AI for obvious humor (like Xfinity and Dunkin) might work, distinctive brands like Coca-Cola risk losing brand equity if they replace human craft with “AI slop.”
Tip for marketers: Authenticity matters. AI-generated creative content faces skepticism, especially for premium brands trying to create distinctive identities.
Watch Dunkin’s “Good Will Dunkin”
Good Will Dunkin’
Humans vs. Machines
OpenAI’s “You Can Just Build Things,” ad focused on human hands sketching, crafting, and teams of people solving problems. Similarly, Google’s ad for its AI tool, Gemini, highlighted a family using the image editor to comfort a child ahead of moving into a new house. This effort to humanize the tools and benefits of massive tech companies, left Arora and Seitz with plenty to discuss.
“Google is just phenomenal with their advertising,” Arora observes. “They’re able to show a use case in a very powerful, emotional way. They’ve always been very good at this.”
This human-centered approach may be particularly important in the American market. “America is the most anxious about AI of any country in the globe,” Seitz points out. “So it is very smart for Google to show the AI doing something that a human wouldn’t really easily do. A mom is showing her child, in minutes, what their future house could look like with all their belongings. In that situation, AI is not necessarily replacing what a human would do.”
Seitz goes on to say, “Telling people AI is ‘almost as good as a human’ is a terrible pitch. The companies getting it right are showing how AI lets you do things you couldn’t do before, not that it does your job almost as well as you.”
The consensus? AI ads are largely message-based right now, and there’s a lot to convey in a short format. Leaning on emotions to convey that story, as Google did, or leaning on the ‘wow’ factor of AI functionality like Co-pilot and Codex, seems to be two dominant approaches for now.
Tip for marketers: Abstract messaging around AI falls flat. Companies succeed when they can show specific, relatable use cases that solve real problems.
Watch Google’s “New Home”
New Home