Inuvo, Inc. (INUV) Earnings Call Transcript & Summary

Investors have been keeping an eye on the different areas of the AI boom. Artificial intelligence's ability to assist with search functions and how it might impact advertising spending revenue is just one of them, particularly in the case of Google, which dominates the world with its search engine, and according to GS Statcounter, in 2022 Google made up 91.9% of the search engine market. Inuvo focuses on being a market leader in using AI for advertising and its Chairman and CEO, Richard Howe, joins us now to discuss.

Richard, thanks for joining us. I was reading about your company, really interesting how you go about your business basically in the ad market. The ad market for online advertising's structurally changing. How are you using AI to take advantage of it?

Yes. The big problem in advertising is, for 30 years, has been using identity as the fundamental basis for targeting, for creative, for figuring out which devices are attached to a person. There's just no aspect of advertising that hasn't been based on identity, and that's going away quickly. Most people would know that as the cookie challenge. And what that does is it actually offers an opportunity, an opportunity to sort of free yourself, if you're a marketer, from marketing that's been based on who someone is, to based on why someone is interested in your product, service or brand. And that's where AI comes in. The technology that exists within AI now allows us to not just solve the privacy problem but really change the paradigm to a more powerful answer to the way you find audiences: why are people interested in my product, service or brand?

You have a really interesting example online I found on YouTube that has to do with the Carhartt jackets and the coats. And could you just kind of break down step-by-step how your company goes about using a product such as this and gaining insights through artificial intelligence without having to glean that information from the users themselves?

Yes, you bet. The Carhartt example is a good one. There's a lot of them, but I'll hit that one since you've asked for that. Fundamentally, this kind of AI is based on a language model. There's a lot of different categories of artificial intelligence; a calculator is AI. But this kind is based on the language. And it's based on the language because, we humans, we experience the world through the language. So it's read the entire Internet. It's being able to associate all the concepts that exist in our language, one to the other and all the pages that are on the Internet one to another. And what that allows this AI to do is to associate things. So Carhartt, take a look at them. As a company and as a brand, they really have sort of owned the worker space, like jackets and outdoor wear specifically for people working hard in the fields. And one of the things that our AI is able to do is associate things that you normally wouldn't think are associated. In this case, with Carhartt, it was the movie Interstellar. So because our AI has read everything about Carhartt and because our AI has read everything about everything, it knows there's a relationship between that movie and Carhartt. Now it knows even more than that. It's not just the movie Interstellar. It knows that Matt McConaughey and Jessica Chastain are actually wearing Carhartt gear in that movie, and as a result there's an audience there that you could target that you would never have been able to identify, simply because the technology doesn't work that way.

So if you're able to read the entire Internet, you're going to come across -- stumble across some things, maybe some connections that you don't want. There's going to be a lot of hidden biases in there, and how do you tune your AI software? And maybe it's not just a concept of tuning, maybe it's more structural. How do you build in safeguards into the AI software to make sure it's coming up with relevant and appropriate terms and linkages?

Yes. So when we have our clients, some of the things, obviously, that they care about is they don't want their ads associated with, let's just say, content that may be not associated with what they're looking for. I mean probably a good example of that is an airline. It certainly doesn't want any other marketing materials on sites talking about plane crashes. There's ways to filter that out. So that's one example, but there are many ways to do that. The good news about using this kind of a wide-based approach and not having any relationship or the use of third-party data, consumer data, in any way is these sort of inherent biases, in the way that we think about biases, don't exist. Because, again, it's not about the people. It's about the reasons why groups of people are interested in something.

Have you done a direct comparison between the old method of using cookies and tracking user behavior across sites and your current method? Is it better or worse? I would imagine just blatantly scooping up everybody's data is going to give you much better results, but maybe I'm wrong.