Much is said in ominous tones about AI, the supposed existential threat it poses to humanity or the prospect that automation in businesses will come at the expense of jobs and the people that perform them.
But Arthur Sadoun, Publicis Groupe's chief executive and chairman, and Carla Serrano, its group chief strategy officer, insist the notion that AI that will take people's jobs need not apply here.
"In our case it's very simple," Sadoun says. "AI is at the service of our people, AI is going to superpower our people, AI is going to help our people grow and be better. And this, by the way, is in a company that is growing and that will continue to grow.
"We have moved roughly in the last seven years from 70,000 people to 1,00,000. AI will need people to work differently and we will train them to work differently."
Speaking to Campaign the day before the launch of an AI-powered entity called CoreAI, into which Publicis is investing $325.9 million (€300 million) over the next three years, and a pre-results peek at February's Q4/full-year results, Sadoun had reason to be bullish.
The market consensus was that Publicis would report 4.3% organic net revenue growth in the quarter and 5.9% for the full year, yet it delivered 5.7% and 6.3% respectively. Meanwhile, following the news, analysts have reacted positively.
The revelation of Publicis' growth and the advent of CoreAI was deliberately packaged into a single announcement. "The new business we've won in the last 18 months is ramping up even faster than we thought because of this need for personalisation at scale," Sadoun says.
According to Serrano, this growing client appetite for exactly that type of personalised work from agencies means not just more AI but a need for more people with AI skills.
"What's going to happen is that people are going to change jobs," she says. "We didn't know what a prompt engineer was a year ago and now every one of our digital agencies has prompt engineers [people who turn language into a form AI can understand] or has upscaled the people within the creative dens into prompt engineers."
Competitive edge
What sets Publicis apart from much of its competition is its early adoption of a "Power of One" strategy (the brainchild of Sadoun's predecessor Maurice Lévy); its agency integration that puts its various businesses on a shared platform; its acquisitions of the likes of consulting arm Sapient and data unit Epsilon; and the launch of AI platform Marcel in 2018, which was rolled out in the UK the following year.

"Other people might have millions of premium assets and performance data points on their people but they're not connected and they're not on a platform that was once controversial, called Marcel," Serrano (pictured above) says.
"We've been basically ingesting data like this in one platform for six years now. If you think about 650 billion impression bids daily, that's because of our media scale, so we know exactly what's performing out there."
Epsilon and Publicis Media make up some 2.3 billion consumer profiles—information about what customers and prospective ones look like.
"The proprietary data we have for CoreAI is already very, very different and it's ours," Serrano adds. "At the end of the day, AI's a levelling factor, we all have partnerships with everyone else so we can definitely enrich our data with some of the other data points that people have."
Publicis Groupe's single platform structure is key to the success and future of AI, the company says.
"AI is available to everybody but it's not mastered equally, because not everyone has proprietary data and the fuel for AI is large amounts of data sets," Serrano says.
"If people [and businesses] aren't connected, what happens is that they can use AI in a vertical or a silo, taking automation and being as efficient as possible vertical by vertical. You hear a lot of organisations talk about that."
However, being connected on a single platform with AI its beating heart means "you can take advantage of AI because you're already inter-connected to innovate and to grow", she says.
Sadoun is more blunt: "The fact we have moved to a single structure, you can have all the data in the world—but if your organisation is segmented, you can't do it."
Structural simplification
Clearly, though, many of Publicis' rivals are setting out to simplify and clarify their operations across their multiple agencies.
Examples are numerous, from the merger of WPP’s VMLY&R and Wunderman Thompson to form VML, Accenture Song’s “compressed and simplified model” to Dentsu’s “One Dentsu” strategy, to name a few.
But Sadoun says adland lacks its six- or seven-year head start and Publicis' scale of proprietary data. "In a world where our competition starts to operate as a platform, then the question becomes one of timing," he says.
"The machine is learning every day and so as we are starting now, our ability to generate more insight, more data, more fluidity thanks to the CoreAI is going to be exponential in the weeks and months to come."
Even if the competition were to arrive at a similar milestone to Publicis within the next 18 months, "it's going to be almost impossible to catch up", he adds.
Wishes for the future
Earlier this month, Publicis emphatically signalled its growing AI intent when it revealed its "Wishes" films, personalised videos sent to all 100,000 staff. But Serrano is quick to point out that, from a creative and staff standpoint, it was far from a case of merely pushing a magic AI button—skilled people are required to operate it.
"When you think about what it requires from a creative concept standpoint to production, to shooting something, to having the technology, you still need people to almost manhandle AI," she says. "You have to direct it. It doesn't have a sense of humour."
In the "Wishes" films, Sadoun's AI twin was marvelled at by the real deal. "You look pretty good... except for the hair," he told his computer-generated self.
It raises an interesting point. Serrano points out that while there was an uncanny aspect to Sadoun's twin, the AI continues to refine itself and improve.
"Even as we were training Arthur's AI twin and his was the AI we trained the most... we watched week by week the model get closer and closer," she says.
"And, even today, if I were to play you his AI twin now, the fidelity is even better than what you saw in some of those films. Give me six more months with his twin and I don't think you'll be able to tell the difference."
Sadoun adds that "no one ever has been able to deliver videos personalised at this level".
"But it's only an experiment and is a great expression of what we are trying to achieve, which is pushing the boundaries of technology to get better work, better insight or better media, but also thinking this will never replace humans," he continues.
"You can’t leverage AI if you haven’t been through the heavy and tough work we have in the organisation, spending billions on technology. We are talking now about $325.9 million (€300 million) but we have spent $8.69 billion (€8 billion) to get there.
"We are starting from a position of strength. And it comes back to what I was saying about not cutting jobs. Because we are making this transformation, we are ready to go even further.
"We are not using AI to fix an efficiency issue. Efficiencies are not what matters. What matters is how we are going to do things better and differently."
(This article first appeared on CampaignLive.com)