Human-Made Ads Still Perform Better Than AI Spots, Study Finds

Audiences can’t always tell when artificial intelligence is behind an ad. That doesn’t necessarily mean they’re as effective.

A new study from a pair of professors at Syracuse University and research firm Ipsos found that consumers were more likely to prefer human-made ads over AI-generated spots, even when they couldn’t tell whether AI was behind an ad or not. While only 25 percent of ad viewers were able to identify an AI-made ad with some confidence, 40 percent couldn’t tell either way.

Regardless, consumers still said that the human-produced spots were more entertaining, unique and conversation-worthy than the AI-produced ads. The business results were even starker: the AI ads were, on average, about 14 percent weaker than the human-produced ads on short-term effectiveness and 17 percent weaker on long-term brand equity. 

The study, which tested 10 human-created ads and 10 AI-generated ads with an audience of 3,000 U.S. consumers using Ipsos’ ad assessment tool Creative|Spark, was geared toward brand marketing and not politics. Still, the research sheds some light on how audiences react to and perceive AI-generated content at a moment when advertisers are racing to adopt AI technology.

It’s not that AI is inherently bad or ineffective, said Adam Peruta, an associate professor at Syracuse University’ S.I. Newshouse School of Public Communications and one of the study’s authors. The AI ads used in the study were based on creative briefs generated from brands’ original ad spots and were deemed “credible” enough to sit in a real media environment.

And the AI-generated ads performed best when the content was product-driven and direct. In other words, when there was a clear problem and solution, the AI ads got the job done. When it came to more nuanced or narrative-driven ads, the human-produced works excelled.

The lesson, according to Peruta, is that ad effectiveness alone can’t be determined by cost, production time or the sheer volume of content produced.

“I think advertisers really have to go back to how they’ve been doing it for decades,” Peruta said. “You’ve got to create a piece of media or a piece of content and you’ve got to test it. Is it going to be persuasive enough? Is it going to cause somebody to take action? Is it helping to build brand equity, conversion – all those things that we care about in advertising?”

One finding that Peruta said stood out: when consumers sampled in the study were told that an ad that they viewed was AI-generated, most said it didn’t affect their opinion of or trust in a brand. What mattered more, according to the study, was transparency; 79 percent of participants said that companies that use AI should have to disclose when it’s used.

It’s also worth noting that generative AI in advertising is still relatively new. With advances in AI technology and increased adoption, Peruta said, AI could stand to become “just another tool” in content creation, like Adobe Photoshop. 

“Given the capabilities of the tools and platforms, we’re only going to see more and more AI-generated content,” Peruta said. “Just my personal hot take: I think we’re going to get to a point where it’s just not going to matter.”