When research group OpenAI first introduced what is arguably the most powerful text generator in the world last year, it immediately wowed the AI community with its ability to produce realistic copy. Trained on nearly 1 trillion words pulled from across the internet, the generator, known as GPT-3, is able to mimic the style of a piece of writing based on something as short as a single passage.
But granting that much creative license to a machine also breeds a certain level of unpredictability in output. That element of randomness is partially why GPT-3—or its less powerful predecessor, GPT-2—is taking time to gain widespread commercial traction as a tool to power chatbots or auto-generate ads.
After nearly a year of experimentation, however, how such a technology might be tamed for marketing purposes is beginning to take shape.
Working through OpenAI’s closely guarded API program, startups and agency technologists have reined in GPT-3’s more eccentric tendencies, which can range from nonsensical prose to inappropriate or explicit content, in order to put it to use for rote performance marketing tasks like A/B testing endless variations of a digital ad, generating product descriptions or assigning email subject lines. Meanwhile, other companies are capitalizing on GPT-3’s stranger side for creative tools.
While still nascent, projects like these offer a glimpse into a future where humans might work hand in hand with generative AI on creative copywriting and the give-and-take forces that might define such a relationship.
We Asked Copysmith’s GPT-3 Tool to Generate Ad Ideas for Different Brands
Brand: Ford
AI ad idea: “What if we did something with a skydiving car chase?”
Brand: Lay’s potato chips
AI ad idea: “What if we performed an experiment with a vending machine that would only accept your shadow as payment?”
Brand: Adweek
AI ad idea: “What if we did something with a giant red button that when pushed, would break some news?”
“In marketing, the trade-off is between guardrails and creativity,” said Gartner Research analyst Andrew Frank. “You can create fairly strict rules that prevent the machines from generating crazy stuff. But I think the real impact is probably more on the fringes, where it’s creating things that people wouldn’t necessarily think of or expanding the scope of language for a brand.”
One of these companies doing just that is startup Copysmith, which recently raised $10 million to support its mission of using GPT-3 to create anything from digital ads, SEO meta tags and product descriptions to blog ideas and listicles. Copysmith users can select the tone of some types of content from a dropdown, with options like neutral, funny or luxurious.
The company, which counts Marshall’s, Volcom and Change.org among its clients, combines OpenAI’s model with its own proprietary algorithms to allow marketers to reword the same message with more guiding control than a GPT-3 alone, according to CEO Shegun Otulana.
“You can produce a whole ton of variations—way more than a human being can—and at a faster rate,” Otulana said. “So if you have a market that you’re trying to reach … then you want to really test a whole ton of variable language and see which one performs the best.”
Dentsu Asia Pacific’s data science division recently released a similar content generation engine built around GPT-3. Users simply input a social media caption, and the program spits back several more iterations designed to be tested on different audiences.
While tools like these take pains to curb GPT-3’s more absurdist impulses, other advertising creatives say studying a generative AI’s random output can offer inspiration. Qasim Munye, founder of GPT-3-based writing assistant ShortlyAI, is harnessing GPT-3’s creative side. His startup is based around a collaborative platform that will suggest additional sentences and ideas as a writer types.
“I wanted to create a tool that will become the next generation of Microsoft Word,” Munye said. “It can help you think through your ideas, it can help you articulate your thoughts.”
As for the next steps for this technology, Sid Bharath, co-founder of a GPT-3 ad and blog copy platform Broca, said he believes it will be the beginning of a synthetic content revolution that will also encompass similar tools for generating AI images and video through text inputs.
“You could, say, generate a video of two people talking and one of them is wearing Nike shoes. And boom—it’s like a little video that you can use in your ads,” Bharath said. “In the future, we’re going to be able to create all sorts of media … using the GPT series.”
This story first appeared in the May 10, 2021, issue of Adweek magazine. Click here to subscribe.
