Pretesting with AI-generated static and video ads

Artificial intelligence (AI) technology has enabled the creation of high-quality ads for pretesting, prior to final human production. In a 'proof of concept' study, AI was employed to produce static and multi-shot video ads. These AI-generated ads were compared to real ads based on them, and no significant differences were found in ad-effectiveness metrics such as recall, recognition, brand choice, brand attitude, and ad liking.

The study was conducted under a low-involvement context with multiple real and AI ads viewed, with all AI-generated ads meticulously curated and professionally screened. The ads focused on factual information rather than emotional experiences. Labeling ads as AI-generated may become mandatory soon, but this research found that disclosing AI-disclosure prior to exposure did not impact the results.

The AI-ad generation and data collection were funded by MediaScience, which also partially funds the researcher's salary. MediaScience and the researcher's institution, the Ehrenberg-Bass Institute, are based in Adelaide, South Australia and University of Bridgeport, Connecticut, USA respectively. S.B. analyzed the data independently and wrote the article, expressing personal views which do not necessarily reflect MediaScience's opinions. The research is licensed under Creative Commons Attribution 4.0 International License.