As we are in the midst of the AI revolution, it makes sense that marketing agencies have raced to create AI leadership roles, and fill them with the most capable minds. With few AI frameworks to base decisions off of, and even fewer proven playbooks for integrating AI into creative workflows, these leaders are tasked with navigating ambiguity while setting ambitious strategies.
In this interview with Peter Van Jaarsveld, Global Head of AI at London-headquartered advertising agency OLIVER, we speak about the considerations of an AI leader and how the role might evolve.
How are you using AI at OLIVER?
It’s almost a question of how aren’t we using it. Since late 2022, we’ve moved AI to the centre of almost everything we’re doing at OLIVER. We’ve had lots of awareness training – one of the first things we did was develop a high-level AI policy course that all staff had to complete. We wanted to get our people comfortable with discussing and exploring AI, while understanding the ethics and impact of using it.
We made sure everyone in the business has at least a bronze level certification in our core AI tools. We have our AI Sandbox, which we’ve set up inside OLIVER that allows our people to use a range of tools. In the Sandbox, we have 11Labs for voice generation, ChatGPT, Anthropic, and a host of tools for client work or other uses.
This includes Pencil, our end-to-end AI creation platform, but its real power is almost as an aggregator of AI. With it, we can swap models out. For example, we can use Google’s models for some accounts, and then OpenAI for other accounts. As models evolve and change, we can alter that. We set up a client council almost on day 1 of AI integration, to help key clients understand how AI can impact them and the work we do with them. Some of our financial services clients adopted AI quicker than expected and now AI is core to the work we do with them.
We’re trying to lean as far as we can into AI.
How do you choose which models to use?
We prefer to use as few models as possible, but the right ones. A big part of our approach is to try and create a wrapper around the models so our users, staff, and clients are interacting with something fairly consistently, regardless of the underlying models. We may swap out those underlying models based on clients we are working with.
When we work with Google, we have to use Google AI models, for example. In other cases, there may be more freedom. We go through a rigorous compliance review on the individual tools and platforms and make sure our clients own the outputs. We have our own creative and technical preferences, but it is a collaborative choice with our clients. We have an end-to-end agreement that once a tool has been cleared to use in Pencil, clients will automatically sign that tool off for use in their company. The most important thing in the choice of models is flexibility, and collaboration with clients.
How do you measure the success of your AI initiatives?
In a lot of cases, we look at media performance. We’ll look at things like ROI or OS to understand how well things are performing. We may look into expanding our reach into smaller segments of a market. It depends, but for most of the brands we work with, the key metric they look for is efficiency when it comes to AI. That can mean either working cheaper or faster while making work perform better. Something like Pencil has tools for predictive performance. So, as we create ads, we can see live how the work is expected to perform. We did a study last year which showed there was about a 40 percent increase in media performance on AI generated assets through Pencil.
What is the most valuable lesson you have learned about AI integration?
You can’t embrace something as profound as AI with a top-down governance approach. I say this because it’s our employees who come up with the AI use cases that we couldn’t have envisaged. For me, AI is not a technology revolution, it’s a people revolution.
The biggest lesson I’ve learnt is that you have to balance control and flow. With AI, there needs to be governance for safe use with as much room for innovation and freedom for your people as possible. I think that creatives should embrace AI rather than fear it.
This is a tool that, in the hands of experienced creatives, allows them to do really incredible things that could not be done before due to budget or timing. We’re learning to help our people do better work with AI and lean into it as a tool for change as opposed to threat.
How will the Head of AI role change in the coming years?
Being an AI Lead is about how you can help people access this superpower and make it useful in their lives. AI leadership is about understanding not only what is cutting edge now, but what will be in six months time, a year’s time, and onwards.
What’s more important is understanding how to create an environment where there’s psychological safety around the use of AI as many people as possible can safely use AI. The technology is important, but there can often be too much of a focus on the tech and not enough focus on what it means for people and the way we work. The primary role of an AI leader is to create a space where employees have an opportunity to fundamentally change the way they work with AI.
How can we fuel better ways of working, rather than powering the ‘sea of sameness’ that AI can create?
