A new study from Massachusetts Institute of Technology has shaken up some of our beliefs about AI-generated content. The research shows that people think advertising content made by AI is better than ad content made by human experts. It seems the AI ads are winning.
This is surprising, seeing the narrative so far has been that using AI makes the content worse. This way of thinking has even been dubbed “algorithm aversion”.
The researchers asked professional content creators and ChatGPT-4 to make advertising content for products like air fryers, projectors, and electric bikes, and campaigns like recycling and eating healthy.
They used four ways of making content with AI and humans: human only, AI only (ChatGPT-4), augmented human (where a human decides with AI help), and augmented AI (where AI decides with human help).
The researchers looked at how people rated the content based on how much they knew about who made it. They had three levels of knowledge: no clue, some clue, and full clue.
The researchers found content made by AI alone or with the final say was more liked than content made by humans alone or with the final say.
They also found telling people who made the content made them like it less, but not enough to change the order of preference.
AI ads are winning. But why?
The researchers think that this could be because people are biased towards humans rather than against AI. Knowing the same content is made by a human expert makes people like it more, but knowing that AI is involved does not make them like it less.
However, people were still willing to pay more for content made by AI alone or with the final say than for content made by human experts or augmented human experts.
One of the authors, Yunhao Zhang,said this study has some limitations.
“Although our research indicates that content produced by AI can be compelling and persuasive, we are not suggesting that AI should completely displace human workers or human oversight,” Zhang said.
“In our research’s contexts, we carefully selected harmless products and campaigns. However, human oversight is still needed to ensure the content produced by AI is appropriate in more sensitive contexts, and that inappropriate or dangerous content is never distributed.”
Report author, Renée Gosline, said, “It took ChatGPT-4 a matter of seconds to produce content on par with or of higher quality than that of the human experts. But it is also clear that the market values human input.”