Nestle and Unilever deploy AI with caution

Advertisers including Nestle and Unilever are some of the big brands experimenting with or simply using AI with the aim of understanding opportunities and pitfalls, among them the difficult issue of copyright on open systems.

What’s going on

Major advertisers have begun using AI in anger, Reuters reports, with cost savings and productivity gains the main benefits. The news service spoke to executives at the Swiss-headquartered multinational food company Nestle, British FMCG giant Unilever, and WPP, which has begun deploying some of the technology across its client base.

Why artificial intelligence experiments matter

AI is a powerful and complex technology that presents enormous productivity gains and cost savings. But the ease of making words and images for brands means that it’s similarly easy for everybody else to make similar words and images.

Copyright on brand assets and materials is critical to protecting a brand’s profile and identity. While some of the more centrally controlled AI language models like OpenAI and Google have a greater ability to put up guard rails, brands need to also think about how they can protect themselves amid the rise of open-source models too, which have proved much more difficult to police.

Despite some gloom, the opportunities for commercial and creative gains are significant, but the issues at play here are much bigger than the bottom line.  

Key concerns

Though most big companies are experimenting to understand the positive potential of this new technology, there are three key concerns that loom large:

• Security
• Copyright
• Bias

What the brands and agencies say

• Nestle global chief marketing officer Aude Gandon tells Reuters that the company is currently working with OpenAI systems, ChatGPT 4 and Dall-E 2, on campaign briefs, which Gandon says can answer “briefs with great ideas and inspiration,” ideas that can then be further developed by human creatives. Recent examples include work for La Laitière brand with WPP’s Ogilvy; it imagined 1000s of iterations of the image that may lie behind Vermeer’s The Milkmaid.
  
  
• Unilever, meanwhile, uses its own AI tech, Reuters reports, and has deployed it to generate written and visual content for sale of its TRESemme brand on Amazon. Copyright, security, and privacy are key concerns for the mass-market company, which is also working to keep stereotypes out of its advertising, whether human or AI-generated.
  
  
• WPP: In an interview with the newswire, CEO Mark Read explains that savings can be 10x or 20x, largely by cutting the production and travel costs. “Rather than flying a film crew down to Africa to shoot a commercial, we've created that virtually.”
  
  The agency group, still the world’s largest by revenue, is speaking openly and positively about its foray into AI, not least at Cannes Lions in which Read appeared alongside NVIDIA CEO Jensen Huang, following the announcement of a deal between the agency holding group and the computing company to collaborate on an AI-enabled content engine.

The evolving issue of copyright

Copyright is probably the most complex, and most uncertain issue that commercially driven AI faces, given that security and bias can be corrected by human workers. The technology is in very early days and the regulatory and legal ramifications are only just coming into view.

Most recently, a US Federal court ruled last week that AI-generated artwork can’t be copyrighted, with Judge Beryl A. Howell arguing in her decision that “human authorship is a bedrock requirement of copyright.” The plaintiffs plan to appeal the decision.

For brands, the difficulty comes when AI systems begin to learn from branded inputs, theoretically opening up IP or information to other users. Some commentators suggest considering every input to an AI as “gossip” that you may or may not want to put out into the world. But such rules illustrate the fact that these remain black boxes whose internal operations we don’t yet understand.  

Sourced from Reuters, The Verge, WARC, Ogilvy