KU Business faculty experts discuss: Artificial intelligence and digital marketing
Generative AI programs like ChatGPT have already become widely used in schools and the business world, but how much do users know about the machines they are putting information into every day? While there are cautions to take when using AI both for work and play, businesses and students can leverage its ability in many ways.
James Reeder, assistant professor of marketing at the KU School of Business, dives into what the public needs to know about generative AI and machine learning and what effects he has already seen on digital marketing as a whole. Reeder’s research focus is on better understanding the effects of various sales promotions and personal selling interventions on business transactions. He is trained in structural modeling, causal inference, machine learning and AI, and uses these methods to answer questions in his research areas. Before joining KU, Reeder worked at the Daniels School of Business at Purdue University and received his doctorate from the Simon School of Business at the University of Rochester.
What does the general public need to know about generative AI and machine learning?
That it’s everywhere. Before, capitalizing on data was something that could only be done by the largest firms with the most money, but I’m finding now that companies of all levels are starting to use these tools. Basically, the democratization of AI is very prevalent and is being used for a variety of marketing tasks, such as dynamic pricing in cases like Wendy’s recent use of surge pricing during peak mealtimes, or how prices of Uber rides fluctuate based on demand.
Years ago, the only companies that did this kind of pricing were airlines because they had a demand system. Now, with access to data and computing power being much cheaper, companies are using data in a variety of ways.
How have you seen the effects of AI and machine learning on digital marketing specifically?
The whole reason why digital marketing exists is because of AI and machine learning.
Companies like Facebook and Amazon for example, capture many points of data about their consumers every time they use their platforms. This allows them to have a deep understanding of their customer base and make more informed decisions overall.
So, having rich access to people’s behavior has made digital marketing much more effective.
Is generative AI and machine learning something that we, as marketers or consumers, should be cautious of?
There are a lot of studies that suggest that rather than offloading your thinking into ChatGPT, if you work in concert with it, you can perform better. There’s a large study from last year that looked at a consulting firm to compare tasks of consultants with and without the help of ChatGPT, and they found that in day-to-day tasks, having ChatGPT leveled the playing field between the good and the poor performers. However, when you got to tasks that were much more novel and strategic, ChatGPT gave very convincing answers that were often inaccurate.
So, there’s a boundary condition.
I caution everybody that you shouldn’t trust what it’s giving you if you’re trying to get facts, because you may be getting links, but those links could be fake. My suggestion for approaching generative AI is to use it as an idea engine that you can verify.
Another thing to be very cautious of as you interact with generative AI, is to be very specific about what the machine learns. Because if you put highly sensitive information into basic AI programs like ChatGPT that are available to the public, the system keeps that information and there are ways that others can extract it back out.
For example, one thing generative AI can seemingly do very well is summarize content, but you should not be putting confidential information into the machines to be summarized.
There are, however, confidential machine learning programs that can be purchased and will reside on your machine rather than the internet, so they can keep your information private.
In what areas of business do you think AI will have the most immediate impact?
Unlike many other technological advancements, this innovation is disrupting white collar work. There are studies that look at freelancing, which is now seeing a nosedive because people and businesses feel less of a need for others to do tasks like writing when they can have generative AI do it for them.
Although, what I tell my students is that if you use any of these programs, you want to at least have some knowledge of the domain areas you’re asking for. Ideation is great if you have the prior knowledge to say what is good or bad.
Academia overall is also facing the issue of how to test students using essays and writing assessments, so now we are rethinking how to evaluate people properly, and I think we will see a rise of pen and paper or oral exams.
There are also professors who are doing very innovative things with AI in academia though, such as training large language models for their classes to be tutors, as studies now are showing that individuals who use AI as a tool to learn, rather than a tool to think are performing much better and learning more in school.
How do you think businesses can balance AI without it getting in the way of authentic messaging?
Again, generative AI is great for ideation if you have knowledge about it and can verify what it’s saying. It is also a powerful tool for thinking through things and problem solving, but if you have a specific customer base, it is probably not in the program’s training set, so businesses have to be wary about if the information from generative AI is sufficient to actually manage what they think their customers will do.
For example, you can have generative AI write you 15 marketing messages, but if you don’t have context for your customer base, then there is a possibility that things will go very wrong. Especially because other companies are doing it, so you run the risk of creating a lot of the same stuff, and in marketing you must find some way to stand out.
What tips do you have for those that are curious about AI?
Just go try it; many of the options out there are free or have a free trial. You will find out really quickly if it’s going to work out or not. Different AI systems do different things well, so you want to have a good sense of each one’s strengths.
The hard part is that at the end of the day, these are just prediction machines, so without additional work it is hard to validate if the content it is creating for you is right or wrong, especially since it may be pretty accurate in some general domain areas.
By Lauryn Zebrowski
