In this episode of the podcast, I speak with Anindya Ghose from NYU and Vilma Todri from Emory University about their recent paper, The Impact of Visual Generative AI on Advertising Effectiveness, which is available in pre-print. In the paper, Anindya, Vilma, and the other authors assess the performance efficacy of three types of ad creative:

1. Creative entirely crafted by humans;
2. Creative crafted by humans but modified by generative AI;
3. Creative entirely created by generative AI.

The authors also quantify the impact of disclosing that creative was produced by generative AI tools. In the paper, the authors run a field experiment (buying display impressions on Google Display Network) as well as a lab study where they collect qualitative data. We discuss the results of these in the podcast.

In addition, we cover:

• The genesis of this paper and the motivation to explore this topic;
• The four research questions the authors pursue in the paper;
• The methodology used to interrogate those questions;
• The principal conclusions from the studies the authors ran;
• The authors' finding that lower creative constraints on the genAI tool led to better results;
• Possible explanations for why a Generative AI tool benefits from total control with the visual modality whereas textual generation performs better with a structured, chain-of-thought process.

Thanks to the sponsors of this week’s episode of the Mobile Dev Memo podcast:

• ⁠Xsolla⁠. With the Xsolla Web Shop, you can create a direct storefront, cut fees down to as low as 5%, and keep players engaged with bundles, rewards, and analytics.
• ⁠INCRMNTAL⁠⁠⁠. True attribution measures incrementality, always on.
• ⁠Universal Ads⁠ is Comcast’s self-serve TV ads platform that lets you launch campaigns in minutes across premium inventory from NBC, Paramount, Warner Bros. Discovery, Roku, and more.

Interested in sponsoring the Mobile Dev Memo podcast? Contact ⁠⁠Marketecture⁠⁠.