A Principled Approach To AI

In news cycles that are driven by attention-grabbing headlines, it can be difficult to separate hype from true shifts in the media landscape.

Take the metaverse, where some of the most anticipated platforms were revealed to garner less than 1,000 daily active users.

However, while the predictions for the influence of AI on every facet of the industry can feel like hyperbole, the data backs it up: generative AI has outpaced both smartphones and tablets in speed of adoption in the United States.

I believe that part of what makes a great marketer is a passion to be on the forefront of creativity, harnessing the latest innovations for clients.



However, it's important to temper this enthusiastic impulse with a framework that ensures organizations are consistent in the way they are approaching nascent technologies.

For agencies, this concerted approach to innovation is all the more imperative. As stewards of our clients' brands, our role is not only to introduce those we work with to the latest developments and tactics, but to ensure that controls are baked in to assess both immediate effectiveness and long-term implications for a brand and its consumers.

To this end, we have has been uniting the brightest minds from across our network -- operational leaders, data scientists, legal and ethics experts, and more -- to develop a meaningful approach to AI for our organization.

We have aligned behind three principles that we believe should be at the foundation of any agency's approach to AI, or nascent technologies on the whole.


We recently surveyed clients across the globe to better understand how they are looking at the future and thinking about the client-agency partnership.

Two directives arose, which can feel as if they are in opposition, but should exist in concert: thinking big and demonstrating efficacy. When assessing how we integrate AI into an organization, both of these imperatives must be taken into account.

An effective approach to AI requires ensuring that we are delivering against our clients' marketing objectives with a clear business return on investment -- going beyond vanity metrics to ensure we are creating relevant narratives that resonate with a brand's most valuable audiences.

We are implementing AI to better fuel our work through insights, building smarter audiences for our brands. We are freeing our teams to spend more time on strategy and execution and less on rote tasks, boosting efficiency by finding opportunities where AI can organize and interrogate data more quickly, leading to deeper and more rapid optimization.

This focus on effective AI has revolutionized how we work and deliver for our clients. We deployed a pioneering solution for activating audiences with a cookie-free, privacy-first approach --leveraging language models that rapidly match tens of thousands of attributes from our internal audience platforms into the activation stream.


Approximately 70,000 AI companies now exist worldwide, with that figure more than doubling since 2017 in the U.S. alone.

With the incredible diversity of companies in this space growing every day, agencies have the daunting task of separating the wheat from the chaff to identify the best-in-class partners that can ultimately identify, develop, and activate the most meaningful media experiences for clients. 

Avoiding a generic media approach is key. Working with multiple technology providers enables the flexibility to tailor an agency's AI offering to the unique needs of each client.

In addition to forming partnerships with the biggest names in this space -- Vertex (Google), Microsoft and OpenAI -- organizations that aspire to be truly agnostic must preserve the flexibility to work with AI companies at all levels of maturity and scale to provide clients with comprehensive service.

For instance, we have worked with Microsoft CoPilot to empower our internal knowledge base with LLM functionality, allowing teams to quickly access synopses of thousands of documents and extract relevant information through a chat interface rather than reading the entire paper.

On the other end of the spectrum, we are working with Faculty, a U.K.-based artificial intelligence (AI) specialist, to give our product-development teams additional machine learning (ML) engineering and modeling expertise to evolve our media toolkit.


In our quest to use AI to deliver for our clients and our employees, we must ensure that people and code work together to deliver human solutions.

Both in the data used to train our models and the output of our AI applications, concerns including privacy, explainability, and bias must be safeguarded through comprehensive governance and an infrastructure framework that is integrated across the organization.

This responsible approach to AI requires alignment from areas including legal & compliance, IT, operational teams, local leadership, and more to ensure a coherent and cross-functional understanding of how the organization engages with AI.

Key concerns include building private gardens that will allow the development of AI applications without leaking IP to public models, creating testing and validation guidelines to ensure quality of content and data before scaling use cases cross-organizationally, and providing access to code and tools to expedite learnings and capabilities.

This approach creates a set of checks and balances, tempering the excitement to experiment within our teams and the governance we need to provide to manage clients' data and IP with compliance.

We have seen some fascinating applications where AI produces some incredible results -- matching products, descriptions, categories for product feeds at huge scale -- only to realize it was making up new very accurate, sensible, and ultimately un-buyable categories.

Supercharging Media Experiences

These three pillars compose what we have termed Meaningful AI -- our ability to dig deeper into data, develop broader and faster insights, deploy more accurately, and operate more quickly to develop better media experiences for our clients.

While each organization must ultimately identify its own dedicated approach AI, starting with these pillars of effectiveness, agnosticism, and responsibility provides a framework that will deliver for clients, employees, and the public.

Next story loading loading..