FILE PHOTO: A view of a Nestle, La Laitiere product cover in a supermarket in Paris, France, August 8, 2023. REUTERS/Stephanie Lecocq/File Photo
LONDON (Reuters) - Some of the world's biggest advertisers, from food giant Nestle to consumer goods multinational Unilever, are experimenting with using generative AI software like ChatGPT and DALL-E to cut costs and increase productivity, executives say.
But many companies remain wary of security and copyright risks as well as the dangers of unintended biases baked into the raw information feeding the software, meaning humans will remain part of the process for the foreseeable future.
Generative artificial intelligence (AI), which can be used to produce content based on past data, has become a buzzword over the past year, capturing the public's imagination and sparking interest across many industries.
Marketing teams hope it will result in cheaper, faster and virtually limitless ways to advertise products.
Investment is already ramping up amid expectations AI could forever alter the way advertisers bring products to market, executives at two top consumer goods companies and the world's biggest ad agency told Reuters.
The technology can be used to create seemingly original text, images, and even computer code, based on training, instead of simply categorizing or identifying data like other AI.
WPP, the world's biggest advertising agency, is working with consumer goods companies including Nestle and Oreo-maker Mondelez to use generative AI in advertising campaigns, its CEO Mark Read said.
"The savings can be 10 or 20 times," Read said in an interview. "Rather than flying a film crew down to Africa to shoot a commercial, we've created that virtually."
In India, WPP worked with Mondelez on an AI-driven Cadbury campaign with Bollywood superstar Shah Rukh Khan, producing ads that 'featured' the actor asking passers-by to shop at 2,000 local stores during Diwali.
Small businesses used a microsite to generate versions of the ads featuring their own store that could be posted on social media and other platforms. Some 130,000 ads were created featuring 2,000 stores which gained 94 million views across YouTube and Facebook, according to WPP.
WPP has "20 young people in their early twenties who are AI apprentices" in London, Read said, and has partnered with the University of Oxford on courses focused on the future of marketing. The "AI for business" diploma offers training in data and AI for client leaders, practitioners, and WPP executives, according to WPP's website.
The team work under AI expert Daniel Hulme who was appointed chief AI officer at WPP two years ago.
"It's much easier to think about all the jobs that will be disrupted than all the jobs that will be created," Read said.
Nestle is also working on ways to use ChatGPT 4.0 and Dall-E 2 to help market its products, Aude Gandon, its Global Chief Marketing Officer and an ex-Google executive, said in an emailed statement.
"The engine is answering campaign briefs with great ideas and inspiration that are fully on brand and on strategy," Gandon said. "The ideas are then further developed by the creative team to ultimately become content that will be produced, for example for our websites."
While lawmakers and philosophers alike still debate whether content produced by generative AI models amounts to anything like human creativity, advertisers have already begun using the technology in their promotional campaigns.
IMAGINED SCENES
In one instance, Dutch gallery Rijksmuseum's research team went viral online on Sept 8, 2022 after using X-Ray to reveal new objects hidden in Baroque artist Johannes Vermeer's oil painting 'The Milkmaid'.
Less than 24 hours later, WPP used OpenAI's generator system DALL-E 2 to "reveal" its own imagined scenes beyond the borders of the painting's frame in a public YouTube ad for Nestle's La Laitière -- or Milkmaid -- yogurt and dairy brand.
Through almost 1,000 iterations, the video of Nestle's version of The Milkmaid generated 700,000 euros ($766,010) of "media value" for the Swiss food giant. Media value is the cost of advertising needed to generate the same public exposure.
WPP said the content cost it nothing to make. A spokesperson for the Rijksmuseum said it had an open data policy for non-copyrighted images, meaning anyone can use its images.
Nestle is not alone in its experiments.
Unilever, which owns more than 400 brands including Dove soap and Ben & Jerry's ice cream, has its own generative AI technology that can write product descriptions for retailers' websites and digital commerce sites, it said.
The company's TRESemmé haircare brand has used its AI content generator for written content and its automation tool for visual content on Amazon.co.uk.
But Unilever is concerned about copyrights, intellectual property, privacy and data, Aaron Rajan, its global vice president of Go To Market Technology, told Reuters.
The company wants to prevent its technology from reproducing human biases, like racial or gender stereotypes, that might be embedded in the data it processes.
"Ensuring that these models when you type in certain terms are coming back with an unstereotyped view of the world is really critical," he said.
Nestle's Gandon told Reuters the company was "keeping security and privacy top-of-mind."
Consumer companies are using data from retailers like Walmart, Carrefour and Kroger to power their AI tools, said Martin Sorrell, executive chair of advertising group S4 Capital and the founder of WPP.
"You've got two buckets of clients: one that is jumping in fully and the other that is saying 'let's experiment'," he said.
Some consumer goods firms remain wary of security risks or copyright breaches, industry executives say.
"If you want a rule of thumb: consider everything you tell an AI service as if it were a really juicy piece of gossip. Would you want it getting out?," said Ben King, VP of customer trust at Okta, a provider of online authentication services.
"Would you want someone else knowing the same sort of thing about you?," he added. "If not, don't tell the AI."
($1 = 0.9138 euros)
(Reporting by Richa Naidu and Martin Coulter; Editing by Matt Scuffham and Daniel Flynn)
