AI in Disguise - How AI-Generated Ads' Visual Cues Shape Consumer Perception and Performance

Abstract

Generative AI’s recent advancements in creating content have offered vast potential to transform the advertising industry. This research investigates the impact of generative AI-enabled visual ad creation on real-world advertising effectiveness. For this purpose, we collaborate with a display ad platform and leverage a quasi-experimental setting that includes over two million ad-day observations by over seven thousand advertisers across nearly 50 product categories, encompassing more than 16 billion ad impressions and 116 million clicks. We find that display ads in which the image was AI-generated outperform ads with human-generated images in terms of click-through rates, but only if these AI-generated images do not 'look like AI.'

We identify key visual features influencing consumers’ perception of an ad looking like it was generated by AI. While AI generates more aesthetic images with larger faces, consumers associate these features with human-made ads. In contrast, in line with consumers’ expectations, intense color saturation in ads signals AI generation to consumers. Our findings have important implications for advertising platforms that offer AI-powered content creation tools and for advertisers adopting these technologies.

Keywords: generative AI, digital marketing, image analytics, advertising effectiveness

JEL Classification: M3, M30, M31, M37