As the development of artificial intelligence produces a constant influx of new use cases, ad agencies are assembling internal organizations to hold discussions on the space and its evolving consequences. These think tanks are not only intended to inform the company’s own efforts with clients but also to help employees navigate new tools and business practices.

Embedded within Omnicom’s corporate structure is a network of roughly 12 committees, made up of senior leaders hand-picked by executives to best represent their area of the business. Each committee is concerned with a different angle of AI strategy. For example, one group is focused on responsible AI, while another is in charge of establishing partnerships with governing bodies shaping the future of the technology.

Also read: How major ad agencies are using AI

Omnicom began assembling the network earlier this year as generative AI quickly accelerated the possibilities for using AI in the advertising world, said Paolo Yuvienco, Omnicom Group’s chief technology officer.

For all of its potential, AI poses numerous uncertainties—from immediate business concerns such as data protection to societal issues that will take several years to manifest, including excessive job replacement and new avenues for spreading misinformation. 

The role of agency think tanks is to consider how these issues will affect the ad industry, and in doing so, form a clear strategy for their business and relevant subsidiaries. The conversations held by Omnicom’s committee on partnerships, for instance, led to its joining the Content Authenticity Initiative, a digital trust group, in June. The company’s committee on responsible AI is likewise engaging with proposed efforts for legislation that will positively affect advertising.

Having had nearly a year to consider generative AI, most big agencies now possess some version of an AI steering group, said Jay Pattisall, VP and principal analyst at Forrester.

How they work

Oliver Agency, which is owned by The Brandtech Group, formalized an AI steering group several weeks ago. Executives sourced representation from all corners of its business, including creative, account management and legal, as well as from clients and technology partners. The group is small—14 current members with plans to never exceed 20—which is meant to avoid pitfalls stemming from an unintended echo chamber, said Peter van Jaarsveld, Oliver’s global head of production. 

Van Jaarsveld equates the group to a think tank because its focus is largely strategic, with the aim of instilling best practices around AI. Meetings begin with an open discussion on current topics in the AI space, before members drill down on particular issues that could affect Oliver and its partners. These include the role that AI could play with respect to diversity, equity and inclusion, legal considerations such as IP infringement and driving collaboration.   

Related: Top 5 DEI marketing efforts to know about

Dentsu also maintains an AI steering group, which is bisected into two teams: one focused on client-facing solutions and another focused on internal use cases. Both teams have fewer than 20 members, meet every two weeks and are overseen by a separate council made up of Dentsu executives, according to Shiva Vannavada, global chief product and technology officer for Dentsu Solutions.

The client-facing team, which Vannavada co-chairs, assesses AI products that have been successful in order to plan new ones. The internal solutions team maps out how best to establish an infrastructure that can be used by employees across the company. Dentsu did not share its process for selecting team members. 

While two-pronged, Dentsu’s steering group often works toward unitary efforts. For example, a partnership with Microsoft, which was coordinated by the internally focused AI team, enabled the development of a generative AI tool called Merkle GenCX, which was led by the client-facing team. (Merkle, the performance marketing agency, is owned by Dentsu.)

Like Dentsu and Oliver’s teams, Omnicom’s committees maintain representation from different corners of the company. Members hail from departments including creative, strategy, legal and finance. Each team is small enough to be fed by two pizzas, Yuvienco said, so that each member is able to have an equal voice in discussions.

And while an executive committee—chaired by Omnicom Group’s chief executive John Wren—oversees the network, individual groups maintain their own leadership. 

Variety and autonomy are important because each committee is dedicated to a different facet of the AI puzzle. In addition to committees focused on responsible AI and partnerships, another team is concerned with findings coming out of academic and research and development circles. A fourth is devoted to Omnicom’s AI-powered marketing and data platform, Omni.

“Diverse thinking is required if you think AI is going to disrupt all areas of business,” Yuvienco said.

How they developed

A quality shared by most of these AI think tanks is they were formulated in the past year. As generative AI rose to mainstream prominence, beginning with ChatGPT last fall, advertising companies realized they needed strategies, and moreover, dedicated groups to design those strategies.

“Clients are asking a lot of questions, so big agencies have to have answers for them,” said Forrester’s Pattisall. 

For example, thanks to an onslaught of AI activity in the absence of industry consensus, the Association of National Advertisers updated its media-buying template to reflect new policies around using AI in client work. The lack of an overarching regulatory framework means that uncertainties extend to many other issues, such as copyright and data privacy.

To be sure, some agencies have managed to entertain these questions through pre-existing internal collectives focused on emerging technology like AI. At Huge, for example, a community of a few hundred employees has been meeting to chat about AI for roughly two years, according to Lisa de Bonis, Huge’s chief product officer.

Casual forums also preceded steering groups at Dentsu and Oliver. The impact made by the rise of generative AI, however, was an incentive for more formal organization, de Bonis told Ad Age.

Dentsu established its two-pronged group roughly four months ago, said Vannavada. The inspiration was noticing how so many of its employees were already experimenting with generative AI tools. 

The seeds for Omnicom’s committees were planted late last year. Yuvienco said that the capabilities of large language models (LLMs) that began to emerge in the fall showed Omnicom how AI could be applied in all areas of its business. Soon after, the company formalized its think tank network to strategize for different departments simultaneously.

As AI becomes more developed, Pattisall expects these groups to shift their focus from abstract thought leadership to concrete execution. This process appears to have already commenced with the erosion of the “play, don’t publish” mindset that was popular six months ago, in which experimentation with tools was encouraged over implementing them in client work. 

But the industry moves fast, and now marketers seem poised to bring to the foreground the ideas they’ve been busily crafting in the background.