Joanne Davis, founder of Joanne Davis Consulting, outlines rules of engagement for agencies and clients to achieve success in using Generative AI, based on a survey of agency and client leaders on how to navigate and activate it.

This article is part of the October 2023 Spotlight US series, ‘Working with Generative AI in America’. Read more

All year, the news has been filled with articles about the good, the bad, and the ugly about Generative AI. From students using Gen AI to prepare their homework, to a lawyer using it to defend against a prosecutor, to content creators using it to develop images. Fashion magazines have even used it to create virtual diverse models – causing backlash for taking jobs away from real diverse models.

Agency and client responses to the AI revolution run the gamut from going “all in” to “don’t touch” due to ownership and other risks. Many clients are thrilled by the speed and amount of content Gen AI can produce, but other clients in highly regulated businesses such as financial services and pharmaceuticals are the ones more likely to have “do not touch” policies.

As Sam Altman, CEO, OpenAI said to the US Congress in May: “As this technology advances, we understand that people are anxious about how it could change the way we live. We are too. If this technology goes wrong, it can go quite wrong.”

In order to ensure that marketers and agencies don’t become one of the examples of Generative AI going quite wrong, it’s imperative that each understand the potential advantages and pitfalls, and map out a strategic plan for how to use these tools.

To get more perspective on this topic, earlier this year my company surveyed 40 agency leaders and four dozen client leaders for input on how to navigate and activate Generative AI, and from those conversations, we developed rules of engagement that agencies and clients can use to move forward successfully with using it. Posed as questions, these need to be robustly discussed by the parties to arrive at a sensible plan and define action steps going forward. We recommend a series of work sessions with marketing, client-side procurement and agency leaders to answer these key questions – but first, let’s look at agency and client perspectives from our interviews (scroll right on the bar below the chart to see all the information):

Agency quotes

Client quotes

“A force multiplier of human productivity and creativity, but not a replacement.”

“Fascinating … we are playing around, no real use yet.”

“Going from zero to three ideas in 30 minutes means the teams can spend more time nurturing the ideas instead of five days figuring out where to start.”

“I like the idea of getting ideas faster.”

“Powerful tools revolutionizing the ways we concept and generate content.”

“Early stages of exploration and some general experiments.”

“Challenges, particularly when it comes to bias, especially when trying to create images in tools like Midjourney and DALL-E. For instance, when we try to get a picture of a doctor, we get only white men.”

“Aware of and sensitive to how it’s used, especially since there can be bias or errors in output.”

“Outputs generated using AI trained on private datasets will be owned by the organization owning the datasets, but when used on public domain data to create content, it might not carry a specific ownership.”

“Talking about it internally with our data and scientific teams – we are risk-averse and taking it slow.”

“Brands and agencies are dealing with thorny issues around who owns work generated by AI, and the answer could be nobody.”

“Legal tends to be conservative so we have tight guidelines.”

“Any content derived from AI tools is only used directionally unless identified as original. We also have policies in place to navigate rights, plagiarism and copyright issues.”

“IT put limits on our ability to access different AI generators, so I’m testing them on my personal devices so I’m ready.”

“The invention of the typewriter didn’t make writing obsolete. The typewriter isn’t the one telling the story, and neither is AI.”

“I’d love to get more ideas from these generative tools faster, but we’re not allowed.”

“Replacing FTEs [full-time employees] with AI is like trying to replace FTEs with untrained 10th grade interns.”

“Very early days but likely will be efficiencies in the long term.”

As these quotes indicate, agencies know more about Generative AI than their client counterparts. A May 2023 survey of 4A’s members showed that over 25% use Generative AI on a regular basis and another 40% have used it occasionally. We believe the percentages have grown since then.

Pie chart showing responses to the question: Have you used Generative AI to create advertising content or assets for internal agency purposes?

As a result, marketers need to take advantage of agencies’ knowledge, and ask questions (though agencies need to ask questions of clients as well).

Marketer questions for agencies

1. How can you:

  • Leverage AI tools to better understand our business? Competitors? Targets?
  • Use AI to enable more value from each FTE (versus reducing FTEs)?
  • Generate more work for us without ownership risk?
  • Get us in market faster?

2. Can you help us:

  • With a strategic plan, versus just testing a bunch of stuff to see if it works – such as, where will our category be in five years?
  • Decide what to do internally? What kind of training? From whom?
  • Understand and leverage potential ROI from AI? What will the new models entail? How can Generative AI help us cut expenses or drive revenue or both?

3. Who is the agency talent running these bots?

  • What new FTE types will be needed? AI trainer, AI ethicist, machine manager, etc.?
  • What creative department training do you have in place, so we know the creative director knows what she or he are doing?

4. What’s your track record?

  • Can you show us relevant cases that used Generative AI and share the objectives, timeline and cost range differences versus not using Generative AI? And results?
  • What’s a reasonable timeline for a certain project from start to finish?
  • Can you provide a playbook for Generative AI?

5. How does the legal protection work with you, and with us?

  • How is your business affairs team trained to manage the rights surrounding Generative AI?
  • How is your account team trained?
  • What’s the escalation process if something goes wrong?
  • What can you share with us to help us educate our corporate leaders?

Correspondingly, marketers should expect and embrace these probing questions from their agencies.

Agency questions for marketers

1. What is your strategy for the use of Generative AI?

  • Do you want to be a leader, a fast follower or a laggard?
  • What is your bravery quotient?

2. What is your governance?

  • Who is participating on a multidisciplinary team?
  • What are differences globally?

3. What are the legal and compliance requirements?

  • How is the marketing team educating legal and compliance departments about safe practices in relation to risk aversion? Where is IT in the compliance loop?

4. What kind of training do you have in place?

  • If your company does not permit the use of Generative AI on company servers, are your people learning on their personal devices?

5. What is your budget?

  • Ultimately, Generative AI has the potential to save, especially because of speed to market, however what are you prepared to invest to learn the ROI of Generative AI?

At the end of the work sessions, you will have a map for your future with Generative AI as you continue to learn.