AI + MKTG IQ > AI MKTG

Our stance on the intersection of marketing and AI

When ChatGPT made its debut in November of 2022, there was an element of novelty to it — almost like picking up a Magic 8 ball, asking a question and having the answer instantly revealed. Sometimes it gave you exactly what you were looking for, sometimes it was off the mark. It could even respond in character, which could be surprisingly spot-on. (Pirate voice, anyone?)

Since then, AI has come a long way. It’s evolving at warp speed, with a constant stream of new tools and platforms. And as we continue to learn more about its potential, one thing is clear:

AI is not a toy. It’s a tool.

Tools make it possible for humans to create amazing things. But they’re most effective in the hands of someone who knows their craft and has the skill and experience to maximize their capabilities.

Here’s an example: Imagine you’re having a computer issue and you call your IT person for help. They look at it, Google a few things and — boom! — problem solved. You may think, “Well, I could have Googled that,” but you may not have found the solution as quickly or easily. That’s because the IT person has the specialized knowledge to not only enter the right inputs but also make sense of the outputs and take the best course of action. They know what to ask and can intelligently judge what they get back.

That’s the basis for our AI philosophy at Anthologic. We believe the best opportunities for brands to elevate their marketing programs with AI pair the power of the technology with responsible human craftsmanship and marketing know-how. And that responsible human craftsmanship should be guided by a written policy with established legal and ethical guardrails.

In other words, we believe AI + MKTG IQ > AI MKTG.

You can’t push a button and skip to the end of the assignment without sacrificing quality and much more. So we think of AI tools like an intern — a resource to help accelerate a step or two by doing tasks that can be time-consuming and tedious, such as collecting and organizing information. Like an intern, AI tools are promising and capable; they just require the right supervision.

Let’s talk about why.

This discussion wouldn’t be complete without at least mentioning some of the inherent challenges that come with working with large language models (LLMs). For example:

• You can’t own what you don’t make. Do you need to have IP rights to your social post? Maybe not. But what about that thought leadership piece or product logo? No U.S. laws currently protect content generated by AI. Only original work done by a human is subject to copyright.
• Copyright is a concern. AI models have been trained on vast amounts of data — including copyrighted text and images as well as content from competitors. So, it’s possible AI may generate content that’s too close for comfort. Just consider these examples from AI testing that produced images nearly identical to copyrighted work.
• Privacy is paramount. Many commercial AI platforms state that they may use the data you provide in your query (prompts, uploads, files, etc.) to improve and train their LLMs. Paid platforms may have measures in place to anonymize the data before using it for training purposes, but it’s important to review their privacy policies carefully — especially when you’re dealing with sensitive or proprietary information. Some brands are already including AI clauses in their NDAs and MSAs for this reason.
• Factual accuracy isn’t guaranteed. LLMs can sometimes output incorrect information, fabricated facts or irrelevant “hallucinations.” Here’s why: Generative AI tools are NOT search engines, nor are they huge repositories of stored data. LLMs operate by predicting the most probable next word, phrase or sentence based on the patterns in their training data. They can produce text that sounds completely plausible, but they can’t distinguish what they legitimately “know” from what they may be making up based on pattern completion. Plus, even though these AI tools have been trained on an enormous amount of data, it’s still only a subset of the world’s information — so they may have blind spots or gaps.

• AI tools can have biases. We don’t know what these LLMs were trained on, so we don’t know where biases might appear or what they might be. But you can almost bet there will be some level of bias in what you get back from generative AI, and it’s important to look out for it. Stock photo sites are one example. You can see some of those visual stereotypes when you conduct image searches (e.g., a smiling mother, father and children in front of a suburban home for “family” or white males in suits for “business”). Generative outputs from a model trained on that kind of data will have those biases built in.
• AI doesn’t always get brand voice and tone right. It’s already harder than ever for brands to connect with buyers, let alone when your brand voice isn’t consistent or authentic. We’ve seen what happens when AI writes brand copy unchecked, and it’s not pretty. Even with a little direction, AI tools can take copy to extremes. “Confident” can yield copy that’s overstuffed with superlatives. “Aggressive” can come out apocalyptic. And human readers will pick up on the difference if your brand doesn’t sound like the brand they know and trust.
• Prompts make a big difference. In our experience, well-crafted inputs are key to getting usable outputs. Giving the AI tool a role, asking it to tackle a solution step by step, and feeding it the data it needs to formulate a response can help — but a knowledgeable marketer can hit the ground running, drawing on experience to create effective prompts and build on the conversation with follow-up questions.

Where we are now.

After 18 months of technology exploration, testing and piloting AI tools here at Anthologic, we’ve learned a lot about what it can do and how it fits in our workflow. Here are two AI plays we’re already passionate about working with clients on:

1. 1. Nimble audience research
      The more you know about your audience, the more you can fine-tune your message to resonate. However, buyer research often gets deprioritized due to limited time, budget and resources. With the right guidance, AI can help you mine the market intelligence and customer data you already have for a relatively modest investment. Although AI tools aren’t a replacement for surveys or focus groups, they can analyze data for audience characteristics that can be validated by sales teams and other stakeholders. These insights can help create audience personas and journeys, buyer profiles, etc., to power more-on-target marketing efforts across tactics — from trade show experiences and social media to email and influencer marketing. AI can also help analyze performance data (for example, email subject lines and open rates or social post engagement) to determine top-performing messaging so you can make adjustments.
   2. Multi-modal content marketing
      Although AI is changing so much, so fast, one thing hasn’t changed at all: Your human audience wants their questions answered, needs met and pain points eased so they can do their jobs better. Now is the time to put a multi-modal content strategy in place and tell your stories with rich, creative, authentic content on land you own: your website. AI tools can assist with tasks from brainstorming topics to transcribing interviews to organizing ideas into outlines in this process — but that’s not the only way AI factors into the equation. AI platforms have an insatiable appetite for premium content across multiple modes, including text, images, video and audio. So, the more multi-modal content you have, the more likely AI platforms will be to find your owned channel answers through SEO, cite your website as an expert source in their generative outputs, and ultimately connect you with buyers.
      
      In addition, consistent, relevant content can help you attract more (and better) customers. And once you welcome them to your first-party database, you can serve them for years to come through marketing automation. This long-game approach is designed to deliver buyers who are more likely to purchase from you, stay loyal and become brand advocates.

What’s next?

Whether you’re an enthusiast or a skeptic, AI is here to stay — and it’s on its way to becoming ubiquitous. When used strategically with smart, focused human oversight, it’s also a powerful tool. Whether you use AI to accelerate your marketing, make it more efficient or tackle things you otherwise wouldn’t have been able to, it can provide the horsepower to help you elevate your marketing program and increase ROI.

Ready to explore how AI can make a difference in your marketing? Connect with us to brainstorm, consult on potential applications or talk pilot programs.

Going to the Iowa Artificial Intelligence Summit for Industry? Anthologic Creative Director Travis Arndorfer will speak about AI-enhanced marketing at the Iowa Artificial Intelligence Summit for Industry on Oct. 10, 2024, at Iowa State University in Ames.

Learn more about the ISU Center for Industrial Research and Service (CIRAS) event here. If you’re looking to get tickets, use promo code AIABI at checkout for a discount! The registration deadline is Friday, Oct. 4, 2024.

Arndorfer previously wrote about Anthologic’s AI + MKTG IQ > AI MKTG philosophy in the August 2024 issue of the Iowa Association of Business and Industry (ABI) Business Monthly. Download the digital edition and check out page 20.