AI Briefing: Creative ways companies used AI in advertising in 2023

According to Trivago CEO Johannes Thomas, the company used to do 35 productions with 20 actors to reach 40 different markets. But that was a “very costly and a long endeavor,” so the company decided to explore how AI might help make “something that is equally punchy” but also would scale.

“It’s not just about dubbing,” Thomas told Digiday. “We can now make the character say different things very easily without reproducing. So we can test scripts, we can tailor scripts to different audiences very quickly and bring this to dozens of markets and scale it very quickly. It’s actually a playing field that just lets us learn faster.”

Trivago is also experimenting with AI in other ways, such as running hundreds of tests each quarter to improve its product experience. It’s also exploring ways to create fully AI-generated characters that can be localized to the look and feel of each market. And to help expedite approval processes like legal checks, Trivago has created an internal “fast pass” with a separate approval track so employees can test free versions of new tools to see what might be worth pursuing. (Employees are also instructed to avoid putting sensitive data into the tools.)

“You need a few front runners that experiment and then do showcases internally to get others excited about about the use cases that are popping up on a constant basis,” Thomas said. “You can even think about different TV shows, different times of the day. If you can make it affordable, you can talk differently to people. So you don’t produce one spot for everybody, but really tailor to the audience.” 

Trivago isn’t alone in experimenting with AI-generated voices. Just in time for the holidays, the meditation and sleep app Calm used AI to recreate the voice of the late James Stewart, the actor who played George Bailey in the Christmas classic, “It’s A Wonderful Life.” 

Using public domain audio samples from Stewart’s old radio dramas from 1940s and 1950s, the AI voice was meant to resemble a version of Jimmy in his late 30s — around the same age he was when he starred in the beloved 1946 film. The voice was developed in collaboration with the Ukrainian AI voice startup Respeecher, which also made an AI-generated version of Darth Vader’s voice in collaboration with actor James Earl Jones and a version of young Luke Skywalker for Disney+ used in “The Mandalorian.”

With “It’s A Wonderful Sleep Story,” the goal was to take a “human-centered approach to AI,” according to Calm Chief Content Officer Greg Justice. In an interview with Digiday, Justice said AI could help “supercharge our content creation engine” and assist Calm as it explores other ways of using AI within the app to create content that wouldn’t otherwise exist. (He also mentioned Calm might also consider using AI to translate its entire English-language content library into other languages.)

“This wasn’t fully tossing the keys to voice-cloning software in chatGPT,” said Justice. “It was a deep collaboration between dialogue and sound editors, creative producers, and the talented team at Respeecher to bring this to life.” 

Here’s just a small sample of how AI helped marketers create compelling content in 2023:

• Orange’s 2023 FIFA Women’s World Cup ad supporting the French national team created a deepfake with AI to address sexism in soccer.
• WWF Deutschland used AI for several compelling campaigns. In one, it recorded humming bees and then replayed the sounds into Google’s AI-powered hum-to-search tool to see which real songs the buzzes sounded like. (The result was 750 songs on “Bee Radio,” which then generated ad revenue to fund the WWF’s wild bee fund.)
• In a separate campaign called “Climate Realism,” WWF took famous landscape paintings and recreated them with AI to generate hundreds of images showing what the same landscapes might look like in 2100 based on climate change predictions.
• Burger King’s “A.I.LLOWEEN” campaign used genAI to create a creepy commercial showing deformed people eating deformed burgers.
• Volkswagen used AI to create a deepfake video of the late Brazilian singer Elis Regina shown driving a VW while singing a duet with her daughter, Grammy-winning singer Maria Rita. (The ad also prompted questions around the ethics of the campaign, echoing the the broader existential ethical dilemma facing the industry.)
• One of the earlier brands to make ads with gen AI in 2023 was Coca-Cola, which used generative AI for its Masterpiece commercial and also used it to create a new platform in collaboration with OpenAI.

Other AI news and announcements:

• Axel Springer and OpenAI announced a partnership to provide news from across the German publishing giant’s portfolio in a summaries format within ChatGPT. (Axel Springer’s subsidiaries include Politico, Business Insider and German publications like Bild and Welt.)

• Google Cloud and Accenture announced a new “joint AI business unit” called the AI Center of Excellence, which will help drive adoption of AI across the companies’ clients. 

• The musician Grimes debuted the first of several new AI plush toys for kids in collaboration with the Curio, a Silicon Valley toy company. The first, named Grok, will be followed by others called Gabbo and Grem. (According to The Washington Post, which first reported the news, the name “Grok” was shortened from “Grocket” since three of Grimes’ kids have Elon Musk as their father and grew up around a lot of rockets.)

• Gartner surveyed marketers for its Generative AI 2024 Planning Survey and found 41% expect “content copilot” uses for AI to have a high impact on productivity, while another 56% said they think it’ll have moderate impact. According to the report, analysts predict organic search traffic results to decrease by 50% by 2028 as consumers shift toward AI-powered search. Even before then, analysts expect 80% of advanced creative roles will be tasked with using GenAI to “achieve differentiated results” by 2026.

• A new report details how a Pro-China network of dozens of YouTube channels used AI to spread misinformation about the U.S. The findings, documents by an Australian security think tank, were first reported by The New York Times.

• A new bill introduced in Washington state aims to promote ethical AI and protect against “algorithmic discrimination.” 

• The Irish Council For Civil Liberties (ICCL) is calling on the European Commission to turn off Big Tech’s “toxic algorithms” by letting users choose what they see in their timelines.

• Analyst Benedict Evans released his annual presentation of predictions for the coming year, which includes plenty of predictions about AI in 2024. 
  
• Finally, the AI Briefing is taking a temporary break over the holidays, and will return on Jan. 8, 2024.