In the age of social media, words wield considerable power, influencing behaviors and shaping societal attitudes. Eugenia Rho, Assistant Professor in the Department of Computer Science at Virginia Tech, and her team are harnessing the potential of artificial intelligence (AI) to delve into the impact of social media rhetoric on public health outcomes, particularly during the COVID-19 pandemic.
Rho’s research endeavors to bridge the gap between online discourse and real-world consequences, shedding light on the interconnectedness of digital dialogue and health trends. By analyzing social media posts from various platforms, Rho’s team identified patterns known as “gists” – cause-and-effect relationships that encapsulate key ideas within discussions.
One notable aspect of their study involved training the chatbot GPT-4 to scrutinize posts from banned subreddit groups opposing COVID-19 prevention measures. Through this analysis, the researchers discerned correlations between online discourse and tangible health outcomes, such as infection rates, hospitalizations, and deaths.
The findings underscored the profound impact of social media rhetoric on public health behaviors, revealing how discussions online can translate into offline decisions with far-reaching implications. By identifying and tracking gists across digital platforms, Rho’s team demonstrated the predictive power of AI in anticipating shifts in health trends and guiding public health communication strategies.
Beyond academia, Rho hopes that this research prompts a reevaluation of social media moderation practices. Instead of resorting to blanket bans or deletions, she advocates for a nuanced approach that fosters dialogue and understanding. Collaborative efforts between social media platforms and public health entities could facilitate more effective engagement with online communities, mitigating the spread of misinformation and promoting informed decision-making during health crises.
As the study prepares to publish its findings, Rho emphasizes the importance of making their methods accessible to other researchers. By sharing their code and insights, they aim to catalyze further exploration into the intersection of AI, social media, and public health – ultimately contributing to better health outcomes and societal well-being.
