Ad agencies can’t stop talking about AI. But how much of that rampant enthusiasm has translated into tangible results? And how are they actually applying the tech? We decided to find out in an effort to separate the reality from the hype.

For many of us, the subject of AI has been inescapable over the past several months. It’s everywhere – whether you’re making casual conversation at a party or doom-scrolling Twitter. At Cannes Lions, temporary seat of the advertising universe, AI was utterly ubiquitous.

But as the hype cycle churns on, many marketing teams seem to be unsure about how to use AI – and especially generative AI – in a manner which is simultaneously productive, safe and likely to stand out from the crowd of many other agencies doing the exact same thing.

As New York Times correspondent Kevin Roose reported in a recent episode of the podcast Hard Fork following his experience at Cannes, “everyone’s talking about AI, but nobody here seems to know exactly what to do with AI. Like, all these companies [are] talking up their AI strategies, and their products, and how they’re using generative AI in experiments for maybe their creative team, but if you really start drilling down into what they’re actually doing, it’s not cutting-edge generative AI stuff… most of it is stuff that frankly existed a few years ago.”

In order to get a clearer sense of the current state of AI within the marketing industry, we reached out to some leading agencies that have been touting their AI credentials with a simple question: how are you currently leveraging AI in order to serve your employees and your clients?

Here’s what we've heard:

VMLY&R
"Until recently, AI required engineers and data scientists. But since ChatGPT, Midjourney and other models exploded onto the scene, the barriers to entry dropped dramatically. So we (and our clients) have shifted from occasionally using AI to determining where to integrate AI into all aspects of our work.

"As an agency, we're embracing the potential of AI while building and deploying as responsibly and safely as possible. That includes:

Responsible AI methodologies: Processes to build humans-in-the-loop; frameworks to mitigate risk; making sure any that we use AI is explainable.

Education and training: Helping our teams and clients understand what’s possible and how to use AI tools and platforms.

AI-Powered campaigns and experiences for clients: Most recently, we launched a generative AI experience with a brand spokesperson.

"It’s an especially crazy time, and as the landscape rapidly changes, we’ll continue to evaluate our strategies, priorities and partners."

- Brian Yamada, chief innovation officer

Code and Theory
"Through a partnership with Oracle, Code and Theory is working to deliver AI-powered solutions to clients in the financial, automotive, retail and hospitality industries.

"Some of the use cases we are pursuing with our clients include:

Implementing a genAI illustration engine for an e-commerce client to consistently refresh their website with new assets, creating a dynamic and engaging user experience.
Building a proprietary research large language model (LLM) for a finance company, leveraging years of proprietary research data to uncover previously unseen trends.
Collaborating with a publisher to envision an AI-equipped newsroom. By utilizing genAI, they can accelerate news cycles, explore new content formats and generate custom content seamlessly, introducing new monetization opportunities and allowing users to personalize their news experiences.
"Internally, we employ AI models such as ChatGPT and Bard for multiple use cases. Apart from copywriting, research sourcing and SEO tagging, our strategy and research teams train these models to mimic client brand personas.

"Our agency has adopted various AI tools and frameworks, supporting our strategy, design, and engineering functions. For image generation, we focus on Midjourney and Stable Diffusion (usually a combination of the two), utilizing a custom private cloud instance through our partnership with Oracle to enhance rendering speed.

"We actively explore new AI tools and encourage sandboxing among our employees through our #AI-Garden Slack channel. Once small teams gravitate towards a specific AI model, generating innovative ideas and concepts, we consider it a viable option and proceed with proof of concept and implementation. Our goal is to continuously improve our workflows by discovering and experimenting with new tools."

- David DiCamillo, chief technology officer

Havas New York
"At Havas New York, we’ve noticed how platforms like ChatGPT and Midjourney have increased our ability to quickly create comps and find supporting research, which leaves our teams with more time to think of big ideas.

"To stay compliant, we've adopted a 'play, don’t publish' approach, which encourages our people to freely explore the strengths and weaknesses of AI with strong guidelines around compliance, confidentiality, attribution and usage.

"Since AI is such a new technology, we encourage experimentation. And our creatives have been sharing tips and tricks with the rest of the agency to help everyone get the most out of each platform. At the network level, we have also established an AI task force dedicated to learning and development.

"AI – while massively impressive – comes with certain limitations, such as the tendency to occasionally produce factually inaccurate texts and to propagate certain biases. So we’re careful not to believe everything ChatGPT tells us. We also believe that AI will never be able to replace individual human experiences and idiosyncrasies, and the influence that these have upon our work."

- Dan Lucey, chief creative officer

WPP
"AI is already a fundamental part of WPP’s business, from automating workflows to accelerating ideation and delivering work for clients like Nike, Nestlé, HSBC and Mondelēz’s Cadbury’s.

"Collaborations with industry-leading tech companies like Adobe, IBM, Microsoft, Google, OpenAI and Nvidia are at the core of our approach to AI. Our exclusive partnership with Nvidia, for example, resulted in the world's first generative AI content engine at scale, as unveiled by Jensen Huang in May. This solution merges clients’ manufacturing data with Nvidia Omniverse to generate photorealistic digital twins of branded products.

"WPP’s data company Choreograph has a central role in our AI strategy, delivering work for clients such as Ford, Unilever, Bayer, The Coca-Cola Company and Verizon. WPP acquired AI company Satalia in 2021, and we also hired our first head of creative AI in 2019.

"Conscious of the potential risks associated with AI, we have introduced measures to ensure responsible and ethical use. The company has established network-wide principles, guidance and legal advice to help our people and clients navigate the potential risks posed by AI, which include biases and misinformation.

"We see AI as an enhancer of human creativity rather than a replacement. The company is focused on learning and development for our people, ensuring that they’re equipped with essential skills for modern marketing – including AI competency."