Toys “R” Us has experienced a lot of criticism following the release of its most recent advert generated using OpenAI’s generative video artificial intelligence tool, Sora.
The advertisement portrays Charles Lazarus, the founder of the American toy retailer, as a child and his aspiration to establish the Toys “R” Us store and its symbol, Geoffrey the Giraffe. Native Foreign, the organization’s in-house studio and creative agency, developed the advertisement.
In a statement accompanying the advertisement, the toy retailer emphasized that its AI-generated film was the first significant advertisement to be created entirely using OpenAI’s text-to-video tool and was “leaping ahead of the curve.”
In a statement, Toys “R” Us chief marketing officer Kim Miller Olko stated, “Charles Lazarus was a visionary who was ahead of his time. We wanted to honor his legacy with a spot that utilized the most advanced technology available.”
Artists and filmmakers objected to the advertisement, with Avengers: Endgame director Joe Russo declaring that it “fucking sucks.”
From a technical perspective, others criticized the advertisement for its lack of visual continuity, which included perceptible changes in the character’s bodily features and clothing throughout the 66-second clip.
One X user wrote, “It is a different person each time you see them, just like a weird dream.”
“This is not the appropriate approach.” Robin Schmidt, CEO of metaverse multimedia firm BasedAF, expressed his confusion regarding the marketing professionals’ belief that this was an appropriate demonstration of their product’s essence.
Others targeted the purported energy expenditures associated with AI-generated promotional material compared to traditional filming.
A former Ubisoft concept artist, RJ Palmer, stated on X that “no aspect of this looks better than conventional tools, yet it costs an order of magnitude more energy to produce.”
“Even then, it is still possible to observe the crude excision of figures from various sources in an attempt to create a cohesive whole,” they noted. “Essentially, there is no advantage to this.”
On February 15, OpenAI introduced its text-to-video model Sora, which immediately captivated users on social media with its capabilities.
Nevertheless, the model’s limitations, such as its incapacity to produce consistent footage without any “uncanny” differences in subjects, have been evident in the rare glimpses of generated footage that have made it into the mainstream since its release.
