Why ChatGPT and other generative AI may help, not eat, your marketing team
/It’s “the fastest-growing consumer application in history” and it’s been out a mere 11 weeks. Whether you’ve only lately heard of ChatGPT or you’ve been asking it to do tricks like explaining in the style of the King James Bible how to remove a PB&J sandwich from a VCR, you may also be wondering somewhere in the more anxious regions of your mind whether and how this artificial intelligence (AI) will affect your work.
I won’t get into the weeds of how ChatGPT works—plenty on that here, here, and here. But knowing the basics will help you understand what it can do. ChatGPT is a very easy-to-use interface built on generative AI, which is completely different from the AI behind Siri, Alexa, and Google autocomplete. Type any question or command into its prompt and ChatGPT will answer in conversational English. It’s usually very fast, often right, and almost always impressive.
But it isn’t really answering your question as much as it’s guessing the most likely, appropriate, conversational response, word by word. ChatGPT is “trained to generate a pattern of words based on patterns of words [it has] seen before.” Having digested billions of words in online documents, webpages, blogs, and social media posts, and other texts, ChatGPT has figured out how to “speak” by seeing how it’s been done before.
ChatGPT does this very, very well—astounding even some computational scientists, who you’d think would be hard to impress.
Every day since last Thanksgiving some new feat by ChatGPT is announced. Yesterday it was a paper in PLOS Digital Health announcing that ChatGPT scored a passing grade on the U.S. Medical Licensing Examination (a standardized test of expert-level knowledge, required for medical licensure in the U.S.)
ChatGPT has been called a toy, and in many ways it is, but maybe like the Altair 8800 computer kits of the 1970s—a toy that quickly led to the personal computer (’70s-’80s), which fed mainstream adoption of the internet (‘90s), which enabled the smartphone (2000s).
To many smart observers, ChatGPT feels like that kind of toy. Scott Lincicome, a sensible, skeptical writer on economics, has played around with ChatGPT and believes that it and other generative AI are “still quite raw but clearly hold tremendous long-term potential for improving, if not revolutionizing, many aspects of our lives.”
It’s clear from Microsoft’s announcement Wednesday of its generative AI-powered Bing search engine and Edge browser (and Google’s rushed intro of its version, Bard, the same day) that most search engines and office productivity software from MS Office to Adobe products will include a generative AI tool, perhaps before this year is out.
This is a speeding train. But is it heading toward us—marketers? How could it all play out?
The faint outlines are already apparent, in the experiences shared by the many writers and coders who are already using ChatGPT (and, lately, AI-enabled MS Edge and Bing). Lincicome quotes Noah Smith and roon (a researcher at a leading AI firm) describing a three-step process they’ve observed in people who are creatively interacting with ChatGPT:
“First, a human has a creative impulse, and gives the AI a prompt. The AI then generates a menu of options. The human then chooses an option, edits it, and adds any touches they like.”
So one way generative AI could play out for marketing copywriters could be something like this writer/AI-coauthor team approach Lincicome is experimenting with:
“I have an idea for an essay; I ask it to write a first draft, with specific instructions regarding style and content; I ask it to refine that draft to correct, remove, or supplement the findings or to add things like citations and numbers; I do this a couple more times; and it ends up producing a sorta-useable first draft of something I might eventually be able to turn into a… blog post (though I haven’t actually gotten to that final stage—yet).”
For graphic designers and artists, it could be something quite the same. The AI will be trained on your firm’s brand guidelines and design grids, best practices on data visualization, and so on. “You’ll write (or speak) a prompt, and the AI will create a bunch of alternative[s],” write Smith and roon. “You’ll then select one of the alternatives and go to work on it,” asking the AI to dive in again with more ideas until you’re at the stage of refining and fixing “little ‘edge cases’ that the AI messes up.”
It’s easy to see marketing strategists producing firsts cuts at ideas and insights with the brainstorming help of generative AI.
The faint outline in these examples is using generative AI to take tasks that are either low-value (competent but unstylish prose) or cognitively burdensome (first drafts) off the plates of marketers and other knowledge workers. But that won’t replace most of these workers. Instead, it will help them focus on other, more value-added things.
“Whatever new technology comes along… it complements our work instead of consuming it,” says Lincicome. In other words, a very possible outcome to generative AI is to free up your marketing team to focus on knowledge work that isn’t so easily replicated by AI: gathering better marketing insights, creating campaigns and initiatives based on those insights (and perhaps helped along in some fashion by AI), and learning from those campaigns by interpreting analytics. A shift to a smarter marketing organization.
To paraphrase Jack Shafer, a media columnist at Politico, marketing doesn’t exist to give marketers a paycheck. It exists to serve their firms’ customers. Generative AI could help us do that, and give us plenty of other, more value-added work to do.
Anyway, that’s the carefully bullish case for generative AI. A few equally smart tech commentators are skeptical that generative AI will change the world, much less our applied creative corner of knowledge work. I’ll provide those viewpoints next week.