Bill Rand Explains How AI Is Reshaping the Marketing Landscape

– By Caroline Barnhill

Go to any marketing conference around the world and you’ll likely be sitting in a breakout session about artificial intelligence and machine learning – and how companies are using these tools to get ahead in the marketplace.

The integration of AI into marketing departments has helped refine their decision-making and revolutionize the way businesses engage with their customers.

We sat down with Bill Rand, McLauchlan Distinguished Professor of Marketing and Analytics and executive director of the Business Analytics Initiative at Poole College of Management, to learn more about how AI is reshaping the marketing landscape – providing businesses with tools and techniques that not only streamline their operations but also deepen their understanding of their consumers. 

How are marketers using AI to gather insights about their customers and produce better content?

Insights into customers and content are often guided by analytics. Before the advent of modern marketing AI approaches, marketers would still use analytics – but these were essentially descriptive analytics where the tools would identify what pieces of content had worked best in the past, or what kinds of purchases customers made in the past. In the last few years, marketing has moved into the realm of predictive analytics where we started to ask questions about what customers would do in the near future, or what pieces of content were likely to do well. However, in all of these scenarios, humans were still generating the actions or the content that the AI was evaluating.

The recent advances in AI have been in the field of generative AI, specifically with respect to what are called Large Language Models (LLMs). These tools are specifically designed to generate content and insights that look like what humans might create. As a result, marketers are now using AI to create content, analyze past customer behavior and prescribe future marketing actions. This has led us to the world of prescriptive analytics where AI can prescribe what content marketers should use. Recent studies have shown that in some cases, AI-generated marketing content in the area of social media can actually outperform human-generated content.

For marketers wary about using AI, what’s a simple way to utilize the technology to ‘test the waters?’

The easiest way is to just start playing around with it. ChatGPT is fairly easy to access… Google’s Bard is now available to everyone… and if you are a Microsoft user, Bing Chat, which is available on the Edge browser, works well. 

Take some task that you do fairly regularly – like writing a social media post about your brand’s new retail location, for instance. Imagine that you are working with an intern and have to give them instructions about what to do. Then give that to one of these AI tools as a “prompt,” which is what we call the instructions that LLMs use. 

When designing these prompts (we sometimes call this prompt engineering), it’s important to think of three key elements: (1) What do you want the AI to pretend to be? For instance, “You are a marketing manager who needs to create content for your brand’s Facebook page.” We sometimes call this the environment or the priming. To add more value, you may want to include additional details about what the brand is and what they sell. (2) What do you want the AI to do? This is the main question of the prompt, “Generate a social media update that talks about the brand’s new location in downtown Raleigh, NC.” You can add additional details about the location such as its operating hours and special features. (3) What do you want the output to look like? This is sometimes called the output or the decorator. “This update should be a short social media post, less than 300 characters, and be formatted so it can be pasted directly into Facebook.”

Remember that these tools are interactive, so if you don’t like your first response, add some more detail and ask the tool to try again.

What pitfalls exist with using AI in marketing that practitioners should be aware of?

There are several pitfalls involved with using these tools, but four that marketers should be most aware of:

• Don’t give away the company secrets – Anything you type into these public AI tools becomes the property of the company hosting it. In ChatGPT’s case, that’s OpenAI, which – despite its name – is not very open and has a large investment from Microsoft. So, do not give any company proprietary information to these tools. It is possible to build a LLM in-house, but that takes a certain level of technical expertise.
• Randomness and hallucinations – These tools are not deterministic, so they may generate different responses even when asking them the same questions. Most of them do feature a “temperature” parameter which can be used to control how random they get, but casual users are often unable to tweak this. This randomness can result in what are called “hallucinations,” where the AI model thinks that certain things are true even though they aren’t. For instance, it could generate citations to papers or books that don’t exist. 
• Limits in knowledge – These tools are trained offline, which means they do not use live internet data. Instead, they are typically limited to data from about a year ago. This allows developers to make sure these tools are at least somewhat accurate before releasing them. It also means that they can evaluate the model before releasing it and do not have to worry about the model changing as people access it.
• Biases and originality – These tools are trained on data from the internet, and since data is biased, the output of these models is often biased as well. Of course, designers are aware of this bias and are working to change it. For instance, if you asked AI to tell you a story about a scientist when these tools were first released, the story would often be about a man. Now the designers have shifted the code so most of these tools will tell you about a woman instead. Essentially, they have incorporated rules to make the AI aware of the gender gap in science. This all being said, the tools are not original. Though the results may look original, it’s because the tools are mashing up content that they have ingested from the internet – and generally, they draw on diverse sources.

What’s the most creative way you’ve seen a company use AI in its marketing efforts?

For most national and international firms, we may not be able to determine if they are using AI to automate social media posts and content generation. There is no reason for them to acknowledge that they are using AI in this space. They might simply start producing more and better content than ever before. 

However, on a small scale, I’ve seen some truly innovative uses of AI. For instance, the company Kimi Visuals, which is an AI animation studio, has done several interesting things, including a case where they took a video of a handpan artist, Johann Immanuel, and had the video switch between real-life and an Avatar-like futuristic worldscape smoothly in a single video. Or the Brazilian artist, Hidreley Diao, who used AI software to create realistic photos of the Simpsons characters. These same techniques could be applied to create amazing marketing content. 

How is Poole College educating the next generation of marketing professionals to utilize AI?

Poole College is devoted to providing our students with the tools they need to face a challenging, constantly evolving business landscape, and that includes the use of AI and analytics tools. All students at Poole College are required to take some analytics courses, and for those who are interested in these tools, we are rolling out classes that dig deeper into these topics. For instance, we already have classes on managing AI projects (MBA 558), managing digital transformations (an MBA 590 class) and implementing machine learning tools (an MBA 590 class).

In addition, we are starting to integrate AI into all of our classes. For instance, in the digital marketing class (MBA 566) that I’m teaching this semester, I’m requiring students to use Generative AI. Specifically, I ask that students use the tool to complete all of their assignments and describe how they used the tool. Looking ahead, I believe that AI will be as commonplace in the workplace as a spreadsheet is today – and we are doing our best to make sure students have the ability to use it going forward. After all, it’s not AI that’s coming for your job – it’s a human who knows how to use AI that’s coming for your job. My hope is that Poole College students will get the education in AI that they need to be successful in this changing world.

This post was originally published in Poole Thought Leadership.