What will AI do to branding? - Rickey J. White, Jr. | RJW™
post-template-default,single,single-post,postid-28541,single-format-standard,ajax_fade,page_not_loaded,,qode-theme-ver-16.3,qode-theme-bridge,wpb-js-composer js-comp-ver-5.4.7,vc_responsive

What will AI do to branding?

What will AI do to branding?

Branded is a weekly column devoted to the intersection of marketing, business, design, and culture.

A few months ago, when artificial-intelligence services that generated images based on text prompts started taking over the Internet, Heinz decided to test just how much its brand was synonymous with ketchup in general.

Working with the agency Rethink, Heinz asked AI image-maker DALL-E 2 to render “ketchup” and variations with different aesthetic modifiers (“renaissance,” “impressionism,” “street art,” etc.). The results, though varied in style, were all unmistakably Heinz-centric—notably referencing the unique shape of the Heinz label. Apart from being a fun stunt, it was a mighty flex, demonstrating the power of the Heinz brand.

But it also hinted at a potential issue for trademark holders that’s lately starting to get more attention: By and large, these AI systems seem to have been built without much regard for intellectual property.

Some observers have been raising this point for a while. Text-to-image tools like DALL-E 2 are “trained” by crawling the Web and “learning” hundreds of millions of text and image associations. Earlier this year, a TechCrunch report noted that DALL-E 2 maker OpenAI “filtered out” pornography and duplicates from that learning process. But the tool can produce images that include logos, trademarked characters, and other intellectual property—such as, SpongeBob shopping at Best Buy, Homer Simpson in Psycho, ancient Rome Spider-Man, Santa shopping on Amazon, or an “angry mob” of Ronald McDonalds protesting working conditions, in the style of Caravaggio.

DALL-E 2 and similar tools created by research lab Midjourney and the open source Stable Diffusion are becoming increasingly accessible—and popular. Video iterations are next, and the quality of the images produced is improving rapidly. “In a few years,” New York Times tech columnist Kevin Roose speculated not long ago, “the vast majority of the photos, videos and text we encounter on the internet could be A.I.-generated.”

As usual, any semblance of a regulatory strategy creating some parameters for managing this technology is trailing well behind these developments. Already, AI-generated images are becoming increasingly commonplace on stock image sites like Shutterstock, iStock, and Adobe Stock. (Getty, meanwhile, has banned AI imagery, citing “unaddressed rights issues.”) Clear answers around authorship and ownership of these images are a work in progress.

For brands, the issues are slightly different, but just as vexing. On one hand, as Heinz demonstrated, AI-generated visuals represent another tool that adventurous companies can play and experiment with. There are a few other examples. Stitch Fix has described how it used the technology to generate images of potential products—to “surface the most informative characteristics of a product in a visual way, ultimately helping stylists find the perfect item that matches what a client has requested in their written feedback.”

Others have examined how AI imagery could fit into marketing campaigns and materials. This speculation has included wondering whether AI could actually become a threat to graphic design or other creative careers.

In the years ahead, AI video could become a radically cheaper alternative to traditional ad production for marketers. “Why spend $100,000 on a single television commercial targeting millions of people,” Brett Winton, director of research at ARK Invest, has argued, “when, with the same budget, an advertiser will be able to create 10,000 different commercials, each tuned to a cluster of like-minded viewers at a moment in time?” Such scenarios may be off in the future, but it’s worth noting that Meta and Google are experimenting with simple AI-generated videos right now.

In the shorter term, AI imagery raises IP and branding challenges that will likely get sorted out in real time. The most obvious case study would probably involve AI creations featuring recognizable IP—a Disney character, for instance, or a trademarked corporate logo. “If a Disney princess is recognizable in an image generated by DALL-E 2, we can safely assume that The Walt Disney Co. will likely assert that the DALL-E 2 image is a derivative work and an infringement of its copyrights on the Disney princess likeness,” a prominent IP attorney told TechCrunch. There could be counter-arguments for transformative or fair use, but IP owners are likely to push the issue as hard as possible. One layer deeper than that is the question of whether allowing these AI systems to “learn” from copyrighted imagery in the first place might be legally problematic.

On a more practical, short-term level, there’s the more workaday matter of how brands and logos and the like might be sucked into grassroots AI experimentation, goofing-off, and meme-making against their will—not for profit, but for lulz. Think of this as the bizarro-world flipside of that Heinz experiment. After all, brands are cultural material, for better or worse, and these are the consequences. In a fairly benign early example, Janelle Shane, on her “machine learning humor blog” AI Weirdness, has already played around with using DALL-E 2 “to mess up corporate logos.” The results are a mix of accidental satire and a kind of critique of any given brand’s relative strength.

It’s not clear that a brand would have an IP-based argument for squelching such material—and it’s even less clear that it would be a good idea to try. DALL-E and its rivals are already incredibly impressive at visualizing whatever we can imagine. But the consequences are going to be very real.

Source: Fast Company

No Comments

Sorry, the comment form is closed at this time.