Criteo, the digital advertising platform, showcased the future of AI in advertising and commerce at a recent event celebrating the fifth anniversary of its AI lab, a centre for innovation in ad technology.
The company, which was founded with the development of a user-based recommendation tool for movies before moving into the field of online advertising, has been specialising in AI pretty much from its inception, according to Diarmuid Gill, Criteo CTO.
“AI is essential to online advertising – it’s impossible to do it without the power of AI,” Gill told delegates.
Gill highlighted the massive step changes in AI including generative AI, such as ChatGPT, which uses algorithms to create new content.
“There’s never been a better time to be involved in AI and Criteo is better placed than anyone else in the industry,” Gill said.
Romain Lerallut, head of Criteo AI Lab, suggested AI was the next industrial revolution. The AI Lab, which was established in 2018 comprises 150 people – 120 engineers and 30 researchers including PhD students. The AI Lab also collaborates with a number of external parties including leading universities and works with product managers and designers.
“The key is making advertising as relevant to people by taking as little information from them as possible,” said Lerallut.
Regulation and AI
A panel featuring Nathalie Laneret, Criteo VP government affairs and public policy; Yann Ferguson from Inria and Adrien Basdevant from the French Digital Council, discussed how to balance innovation and the regulation of AI. Hosted by Live Ralaivola, head of AI research at Criteo, the panel explored how the upcoming AI act could impact the future of advertising.
Laneret said Criteo was interested in three specifics of the new rules: that it adopted a risk-based approach and that online advertising would not fall under high risk use cases; how the supply chain would be regulated and that open source developers will be exempted under the Act.
Ferguson said there was an AI relationship paradox. Rules normally create more trust with a technology but focusing on the potential risks of AI has increased the level of fear.
Management are keen to introduce AI in the workplace because they see it as a good opportunity to enable employees to focus on higher value tasks but workers don’t want it, Ferguson told delegates.
However, with AI like ChatGPT, which has arrived via the internet versus top management, users like it and find it interesting. As a result, there is a mindset gap with management warning employees to be careful of AI like ChatGPT but also telling them not be afraid of AI, Ferguson said.
Basdevant emphasised the need for both regulation and innovation and said the European Commission was striving to keep that in mind. However, he added that generative AI, which has been trained on vast quantities of data, had not been considered in the original Act since the technology was not in the public domain at the time. The European AI Act has no impact assessment for generative AI but regulators would be trying to understand that in the upcoming trial, he said.
How regulators are going to implement the rules and balance the benefits and risks will be key, said Laneret. There are three likely options:
- A brand new regulator in charge of AI for all sectors
- Entrusting regulation to data protection authorities
- A network of different regulators assessing benefits versus the risks, which was the preferred approach
The panel agreed the impact of the AI Act would extend beyond Europe in the same way GDPR applied to non-EU actors.
Advertising and AI
Marc Fischli, executive MD, EMEA at Criteo and Mathieu Morgensztern, CEO of WPP and GroupM France leader, discussed the influence of AI in business and specifically advertising.
Fischli said the industry needed to put the fun back into advertising and AI created an opportunity to do that.
Morgensztern said AI was totally changing the advertising and media industry and a global leader like WPP needed to keep ahead of the game. He went on to share his observations on the role of AI, adding: “Fun is the best way to be relevant and everything about AI is relevance.”
Morgensztern shared WPP’s award-winning AI-led campaign for Cadbury in India. The campaign aimed to empower 20,000 small retailers to advertise in the name of their own store. The challenge was to execute at scale. The campaign brought together a well-known Bollywood star and generative AI to enable store owners to write their own prompt and have the actor star in their advertising. According to Morgensztern, it put a massive generative AI campaign in the hands of small retailers.
“It was totally unexpected to see the star talking about a small store in a region of India,” he said.
In a second initiative, WPP used AI to create live audio commentary of tennis matches at leading tournaments including the US Open and Wimbledon for blind people. The technology monitored the positions of the ball and player using sound and enabled users to have a full tennis game experience in real time thanks to AI. “The emotion felt by the audience was incredible,” said Morgensztern.
Morgensztern went on to share how WPP is driving creativity with generative AI through three key partnerships. The company is working with Getty Images and using a generative AI solution to generate new images based on Getty Images’ stock. It is working with NVIDIA to develop generative AI-enabled digital content such as 3D assets for advertisements and is using content from Adobe and generative AI tools to create new images and introduce them into advertising scenes.
AI is also critical for retail media, the fastest growing sector. “We have a strong ability to predict the outcome of a campaign because we have the data,” Morgensztern said.
Making lives better
Todd Parsons, CPO at Criteo, showed how AI makes lives better and is a force for good.
“Our purpose in using AI is very focused and that’s deliberate – everyone wants to delight a human being,” said Parsons.
Parsons presented a typical shopper journey beginning with an online search about cycling safely in the winter at night and showcased how AI worked across multiple levels to produce results for, not only the shopper, but the publisher and retailer.
AI delivers relevant copy about cycling safely at night in winter for the shopper, it generates side bar ads for the publisher, featuring product recommendations; while generative AI contextualises those ads with snowy backdrops. When the shopper moves through to the retailer’s website and engages with a chatbot, they are presented with further product recommendations, sponsored product offers and new brands.
“Criteo is based on making great recommendations on the buying journey,” Parsons explained.
“AI links all interactions. It is adding value to discovery for the shopper and making them feel safe and dressed for success; the content drives higher advertising value for the publisher and the retailer is able to introduce new brands and drive sales. AI is emphasising the performance to each of those partners,” he said.
Live demonstrations of AI powered tools
Criteo’s AI Lab showed how it is changing the face of commerce through AI with live demos. The AI Lab showcased how AI can be used to improve search results with product attribute extraction on products like TVs, for instance; and for filtering content, such as weapons and adult material, through an ad safety tool. Researchers also shared how AI can help identify audiences and not just be deployed for ad placement.
The AI Lab showed how AI provides enhanced product search too, using both textual and visual cues to locate the most relevant products and showcase these items alongside sponsored ads, for example. The AI Lab also showed how generative AI can help brands create better banners and make images more appealing. Generative AI enables advertisers to change the backdrops to their products, be it furniture or cars, to present different settings and use cases but without incurring the costs of taking multiple products to shoot at different locations.