The following Q&A was edited for length and clarity.

How does AI compare to other digital innovation trends that have surfaced over the past few years?

Unlike the metaverse or some other trends, the power of AI, what it can do, and what companies need to do to harness it is real.

It is a big change, particularly in the way that people can communicate with computers. You can ask a computer to do something, and the its ability to understand you and then do it is a quantum leap from the clipped phrases that we stick into Google.

What does SmartAsset do?

Our SmartAsset product takes a piece of content and, as I like to say, puts it through a car wash. It breaks it down and sees that there's [for example] a cat, a dog, a baby. It understands those elements and gets test results on how the content works.

The biggest danger is if clients don't use what we call "private ChatGPT" then they will be giving out their information and data into the larger pool. We're giving clients a kind of moat around which AI can be utilized.

We have to build private spaces that utilize their data to give them unique understanding and results, otherwise they'll just get the same thing that everybody else gets.

How are you using AI at Stagwell?

Internally we were looking at processes that can be sped up or made more efficient.

When I started working on surveys, I needed about 63 people to do a survey. Now we can do the same project with two or three people.

We're creating efficient products for use in communications, research, and media. PRofet is one — it uses generative AI to help people create a news release, figure out who's going to cover it favorably, and then write the pitches. It's a very useful product.

We also have research products that help people do their own surveys and use AI to analyze focus groups and find statistically relevant findings and articulate them. The research community laps up technology very quickly.

AI will make image creation, testing, and analysis fundamentally more efficient, in the same ways that I saw my survey business become more efficient every three or four years.

How do you foster innovation and adoption of new technologies internally?

I created the Stagwell Marketing Cloud group, and one of the differentiating elements here was to have innovation at scale.

We have a private AI environment internally. So right now all of our agencies come forward and say, "I think I could use AI for X". We take their idea, and we decide which ones we're going to program or engineer for them. About 250 people internally are signed up against our central, internal AI engines.

It's a process for helping each of the agencies come up with ideas and manage the deployment. At some of the [advertising] holding companies, everyone sort of does their own thing in the corner. We really try to get everybody on the central platform.

Do you think the marketing sector is particularly vulnerable to AI ultimately doing the jobs currently done by humans?

I don't think in the pure advertising space there's anxiety about it.

If you look at advertising itself and what it takes to really generate tier-one [premium] content, I always say AI is having an entire group of C students. You can get a lot done with C students, but there's still a whole bunch of stuff you'll never get done.

I do think that the tier-one content could be facilitated, with some assistance to make it easier. Just as processing data became so much easier for surveys when I was doing that.

I think tier-two [stock] content has a lot more danger in terms of the ability to have somebody create it. There is going to be real dislocation over time in that sector.

How do you counsel clients that want to get started in AI, but are intimidated?

Pick off bite-sized projects you can do as a demonstration. Take your last year of focus groups, run it through AI, and see the analysis and recommendations produced through that process.

It's about taking a need and narrowing it down to something practical, rather than trying to boil the ocean.

The biggest thing consumer companies should look at is how to change the interaction between your company and the consumer, through a natural language system.

Say you're an office supply company. On its website, you could just say, "I want to have an office party." The website will ask, how many people you're having, what time, any holidays? And then from that it'll tell you all of the things you need to buy, and which are the top rated things.

That's the kind of interaction where it greatly simplifies the tasks that you're doing.

How will your company interact more conveniently, and holistically with the consumer through large language models that can translate words into action in a much more effective way? I don't think companies put as much time into that.