When IPG-owned performance marketing agency Reprise Digital began using a form of the same AI that powers ChatGPT a couple of years ago, it started with a test.
The firm pitted its OpenAI-powered copywriting software, Transcribe—paired with human editors—against internal and external teams of copywriters to write digital content for financial clients. Each team’s results were judged on factors like cost and turnaround time.
“We saw that Transcribe was magnitudes more efficient when used by people that are changing its output a little bit versus external copywriters or internal teams,” said Vincent Spruyt, Reprise’s head of AI. The tests clocked an 87% reduction in production time between the AI and internal copywriting teams.
These results might confirm some people’s fears—that AI is coming for our jobs—but the reality is more nuanced.
As agencies and marketers begin to implement much-hyped AI language tools like ChatGPT, they are likely to change their priorities for marketing jobs. Experts expect a shift to quality assurance and editing roles rather than pure creative when it comes to performance marketing. At the same time, brands and agencies predict that strategy jobs will become more valuable as organizations zero in on roles that can help direct how AI will be used.
At Reprise, Transcribe has allowed copywriters to focus more on strategy and keyword optimization rather than agonizing over every word in a given chunk of text, Spruyt said.
As AI tools are further implemented across the production chain, to the point where performance gauges like social listening can feed directly into production software like Transcribe, the makeup of teams could shift to focus more on overarching strategy. “That’s how the future of dynamic creative optimization is going to look,” Spruyt said.
That future may not be far off. A recent survey of 1,000 business leaders currently using ChatGPT from job site ResumeBuilder and market research firm Pollfish found that around 58% were exploring use cases for marketing and content. Around a quarter said the system has saved their company at least $75,000 via more efficient processes in areas like writing code and content, as well as copywriting and customer support.
“I am surprised at how quickly it’s being accepted and applied,” said ResumeBuilder chief career adviser Stacie Haller.
To prompt or not to prompt
While David Stein, vp of performance marketing at Wunderman Thompson, has already seen marketing or agency job postings soliciting candidates who can understand the esoteric lexicon of prompt language that has evolved to coax the best results out of AI tools, he thinks such hiring is ultimately short-sighted.
Given the breakneck pace of this new technology, it’s only a matter of time before these skills are rendered null by intuitive interfaces or easy-to-follow scripts. “Eventually, AI is going to get smarter and more intuitive,” Stein said.
Wunderman sent a memo to clients in February warning them off of AI-generated content for now. It cited the technology’s potential for inaccuracies and the embarrassing scandal that played out around tech publisher CNET’s error-riddled AI-generated posts in January.
“We’re not in the ‘oops’ business,” said Stein. They aren’t the only ones: At Reprise, Spruyt said some of the agency’s big brand clients have requested that the AI not be used for their brand.
Demand for human editors
Given the propensity for inaccuracy, most experts agree that language AI implementation will require more quality assurance staff who can check for factual errors and instances where AI may veer off into what machine learning scientists call hallucinations, or meandering non sequiturs unrelated to a given prompt.
Stein said there will likely also be more need for subject-matter experts who can take a machine-produced blog post and add the human experience and knowledge to make it useful and informative.
Of course, in this stage of adoption, a working knowledge of how tools like ChatGPT operate and how they can be used for creative purposes is already in high demand.
At AKQA, which has become known for creative AI work like a recent commercial for GoFundMe, the AI boom is pushing the agency to beef up teams that understand strategy around how to best deploy these tools, said AKQA North America president Tesa Aragones.
“We’re hiring people to help us build teams in AI in technology and data practice,” Aragones said. “So overall, using the technology and the data actually helps us to be more creative and imagine how we can do more for our clients.”
This story first appeared in the March 2023 issue of Adweek magazine. Click here to subscribe.
