From tools to use to chatbot challenges, marketers share how they’re navigating AI

AI, AI, AI: It’s the hot new buzzword on every marketer’s mind, but does anybody actually know what they’re talking about? (We’re assuming, dear reader, that you’re not Sam Altman.)

While Altman didn’t appear at Marketing Brew’s event, The Marketer’s (Early) Guide to AI, we did speak onstage with a number of agency and brand executives to hear how they are incorporating generative AI in their marketing.

Read some excerpts from the onstage conversations below.

Nigel Storey, CTO of Zappos, on which AI tools to have in your arsenal:

My guess is everyone here is probably already using ChatGPT, but ChatGPT, Gemini, Perplexity are the primary three that everyone should be testing, playing with, seeing how they work. I think Perplexity is pretty clever and cool in the sense that it cites the sources.

Michael Cohen, global chief data and analytics officer, Plus Company, on crafting safety guardrails around generative AI:

We have an AI community practice. Within the community practice is a policy group set up by myself and our general counsel. That’s the team that needs to come together. When I think about the two things that we need to really keep a sharp eye on for agencies, it’s the data that’s going into the AI and the license agreements that the software has. The first thing that we lead with is, Do we feel protected with the license agreements for a particular software?

Daisy Jones, agency solutions manager, Meta, on navigating challenges with chatbots:

Instagram Direct, Facebook Messenger, and WhatsApp all have unique restrictions and data privacy policies. Within WhatsApp, because it’s tied to a user’s phone number, as long as there is opt-in to messaging or texting, there is the possibility to use that audience that a client might have already gathered through SMS types of opt-ins and things like that. For Instagram and Facebook, that opt-in has to happen directly in the platform.

Jennifer Kattula, general manager of global marketing, Microsoft Advertising, on training AI:

We actually did an experiment in between when we were prepping where we trained the AI with context: Here’s our brand personality for Microsoft Advertising; here are the attributes of that personality; here are the descriptions of those attributes: Microsoft’s brand personality as a trailblazer is kind, daring, and deep. Then we took that as the context and said, rewrite this blog with that information. Rewrite these social posts with that information.

What we found was that the [Microsoft] Copilot was helpful in some cases, [and] sometimes too literal. We didn’t do any reprompting. I think that’s where the patience comes in: To help the Copilot learn some of the nuance in terms of being too literal. Sometimes it would say, you can be daring, when you really want the essence of daring…What we’re going to try next is starting with some more raw content. Not a fully formed blog that was already written by somebody that really understands our brand, but maybe a set of bullet points. Then have that get to 80% with Copilot and then have a human refine it to the last 20%.

Devika Bulchandani, global CEO, Ogilvy, on using AI without fearing it:

Every single time we’ve prognosticated that something’s going to come and kill us, including Covid, it didn’t. We prevailed...By the way, we create most of these things. It’s not like somebody else from Mars came and created it.