Developers and funders lack evidence on whether AI-assisted health content informs audiences better than non-AI-assisted posts on social media.

Content creators on social media ​​play an important role in shaping public understanding of health for audiences who increasingly look to digital platforms for information. However, the accuracy and effectiveness of ​creator​-generated health content can vary widely.

A​kari​ is an AI-powered ​platform​ designed to help ​creators​ produce accurate, engaging health content. While such ​platforms​ ​offer structured guidance to improve content quality and alignment with evidence, the real-world impact of the content on audience knowledge, attitudes, and beliefs has not been tested. Science ​t​o People initiated this evaluation to generate evidence on whether Akari’s AI-assisted ​health content is more effective than traditional, non-AI-assisted approaches.

NORC designed a randomized study with a nationally representative sample, comparing content made using Akari to content made without AI assistance.

NORC partnered with Science ​t​o People to design and implement the evaluation. We advised on the theoretical model guiding the evaluation and developed a research plan.

We fielded a survey on AmeriSpeak^®, NORC’s probability-based panel designed to represent the U.S. household population, reaching approximately 1,000 respondents. Respondents completed baseline measures of health knowledge, attitudes, and beliefs before being randomly assigned to view either A​kari​-assisted content or content created without AI. Follow-up questions measured changes after exposure to the content and differences between exposure groups.