As marketers rush to integrate AI into their workflows, they’re sorting out what’s truly useful, what’s hype, and how much human input to retain.

“The lesson we learned the hard way is that these underlying language models underneath many applications, even the ones we build ourselves, are not always up to date,” said A. Lee Judge, cofounder and CMO of Content Monsta during “Zero to Launch: Using AI for Campaign and Content Creation at Scale,” a session at the MarTech Conference earlier this month.

Here’s how marketers are weaving automation and LLMs into their creative and operational processes.

Testing tools and standing out

Simply using AI is no longer an advantage, and marketers must now differentiate through how they use it.

“AI is relatively novel, and if you’re using it, you have somewhat of a competitive advantage, but it will be ubiquitous,” said Eric Mayhew, cofounder, president and chief product officer at Fluency. “We need to think about how we differentiate ourselves in the use of AI.”

That starts with scrutinizing tool quality, said Judge. He learned this firsthand when a writing app resurfaced old content, later finding out it was built on an old version of ChatGPT.

“We go straight to the LLM and keep an eye on the latest model,” he said. “We’d rather connect directly to OpenAI than use a third-party app that’s just a graphic interface.”

AI content is losing its shine among consumers, forcing marketers to use it carefully. In 2023, 60% of consumers said they prefer genAI creator content to traditional content, that number dropped to 26% this year, according to a Billion Dollar Boy survey.

Using live experiences and anecdotes as the base of AI will lead to originality and competitive edge, said Judge.

“Start with a human, whether it's a conversation, an experience, meeting notes, or even a sales call, and then use AI to expand and scale that information,” he said. “If you don’t, you’re more likely to be repeating something somebody else said.”

Managing risk through education

Education is often the bridge to buy-in, said Angela Vega, director of capabilities and operations at Expedia Group.

• Lack of knowledge and skills (39%) and integration challenges (35%) are the top barriers holding marketers back from AI adoption, according to a February WARC and Intuit Mailchimp survey.

“A lot of the reasons why internal red tape exists is because IT, HR, and legal are thinking about what happens if a machine gets something wrong,” she said. “Companies that have a strong AI strategy communicated all the way down to the people doing the work see much more desire to adopt it.”

Implementing AI responsibly means identifying tolerable mistakes, said Mayhew.

“There are things you can and can’t accept for risk,” he said. “Go through your processes and find where you have some tolerance. Those become your early opportunities.”

Protecting brand voice

To avoid generic output and retain brand identity, some marketers are scaling AI with segmented, client-specific data. “We build every client’s personality portfolio first, and that is isolated for that client in their own folder,” said Judge. “Whatever we create for them is in their voice, and it's not contaminated by other content from the world or from other clients.”

Some 90% of marketers think of operational costs and time savings as a benefit to AI in analytics, according to a September SAS survey, but marketers that haven’t standardized their processes could be falling short. Expedia Group is testing a “model context protocol,” which allows LLMs to reference specific data sets, like brand standards or performance metrics.

Mayhew, who has transitioned from treating AI like foolproof software to a “coworker counterpart,” said the 10% of the time AI messes up shouldn’t negate the 90% of times it’s helpful.

Feeding AI the right amount of information is key, said Vega.

“Sometimes we think if we dump in all this stuff, that's going to make it more useful. That's not necessarily true,” she said. “As you’re starting to understand how to use AI successfully, it's finding the right amount of context. “

This article was prepared with the assistance of generative AI tools to support content organization, summarization, and drafting. All AI-generated contributions have been reviewed, fact-checked, and verified for accuracy and originality by EMARKETER editors. Any recommendations reflect EMARKETER’s research and human judgment.