The rejection of the new Coca-Cola’s AI Christmas ad It has been basically unanimous: although their bet has technically and visually improved the very promising 2024 bet, the complaints are no longer about the theoretical as much as about something more intangible. An advertisement that appeals to traditional and artisanal things should not be made with a tool that ignores human creativity. Or has it not been like that? Some recent data that has come to light after the first negative reactions casts doubt on whether it was just a matter of pressing a button for an ad to appear.
The announcements so far. In November 2024, Coca-Cola became the internet’s quintessential corporate villain. Its Christmas ad was recreated with artificial intelligence the iconic ‘Holidays Are Coming’ spot from 1995and was received with a wave of criticism, especially for its multiple errors: rigid truck wheels, faces of people frozen in an inhuman rictus… Artists such as Alex Hirsch, creator of Gravity Falls, dedicated to the corporation such strong phrases like “Coca-Cola is ‘red’ because it’s made with the blood of out-of-work artists.” A year later, Coca-Cola does not back down, but launches a new version of the advertisement, more technically advanced and starring less risky entities: anthropomorphized animals.
Pratik Thakar’s quoteglobal vice president and head of generative AI at Coca-Cola, “the genie is out of the bottle and no one is going to put it back in,” has become a symbol of these new times. No matter how much these types of decisions are criticized, the savings in energy and personnel are so significant that these types of changes are here to stay. Or not?
Less effort? There are some figures, made public by Coca Cola itself and by Jason Zada of Secret Level (the company that developed the ad) that cast doubt on the effectiveness of the entire effort. For example, it took approximately 70,000 AI-generated video clips before arriving at the final result of 60 seconds. Behind them, an army of professionals: approximately 100 people distributed between Coca-Cola, the WPP agency, and the Silverside AI and Secret Level studios. And among them, at least 5 AI specialists worked specifically on technical refinement and content generation.
Zada talks about a direct team of 20 people dedicated to this announcement. It is not a revolution in efficiency, but an amount comparable to that of any traditional animation spot. The difference: no physical equipment, locations or cameras were needed, but all the usual production apparatus in this type of ads: creative direction, design, narrative construction, artistic supervision… Zada states that there is “a lot of human craftsmanship involved. Hand-drawn character designs, world-building… it’s not just about writing words and pressing buttons.”
The paradox of money. If we talk about video generation tools like Sora, Runway or similar, each clip has an associated cost. Multiplied by 70,000 generations, the expenditure on server infrastructure, processing and rendering time reaches considerable figures. To this we must add the cost of a hundred people working for approximately a month (this is what Coca-Cola claims is an advantage in terms of time, compared to the several months that traditional production would require). We don’t know how much the ad cost, but Manolo Arroyo, the company’s marketing director, is limited to stating which was “cheaper and faster than traditional methods”
But the important thing here, perhaps, is not how much we save in money, but… is that saving worth the reputational cost?
What is the difference. The type of work, not the amount. Where a traditional studio would spend weeks on 3D modeling and animation, this project invested that time in a process of mass video generation, selection and refinement. Instead of building a 3D model of a herd of seals and animating them, the team generated thousands of different versions of seals until they found the ones that worked. And then, and here is the key, it is able to multiply the result. Zada says, “We could create a 90-second version in addition to the 60-second spot, and a custom version. We couldn’t do that without the efficiencies of AI.” That’s the secret: not to do the same thing cheaper, but to do more things with the same budget.
Coca-Cola doesn’t save money, it redistributes it. Instead of one definitive version of the ad, they got multiple versions tailored to different markets. Instead of investing in filming equipment and physical locations, they invested in management capacity and immediate multiplication: the industrialization of factories that we experienced at the beginning of the last century is now the industrialization of content.
The genie in the bottle. Generative artificial intelligence is already part of the daily life of audiovisual production. And the controversial Coca-Cola ad exemplifies what companies want to get out of this new situation: it is not just about greater speed, or saving money, something that we already see is not being achieved, but rather a commitment to the future, perhaps to a different economic model, perhaps to some spots that are still to come, indistinguishable from those made in a traditional way, and that do not unleash the pejorative comments that, for the moment, these ads made with AI are collecting.
In Xataka | The “divorce” between Coca-Cola and Nestlé leaves a big question: who owns the “formula” of the soft drink
GIPHY App Key not set. Please check settings