Exploring Gymflation with AI (cbowdon.github.io)

A recent project titled "Exploring Gymflation with AI" investigates the phenomenon of increasing gym performance standards on social media, termed "gymflation." By leveraging the YouTube API, the study aims to analyze trends in strength feats shared online, specifically focusing on deadlift achievements. The project employs R programming and various packages to systematically collect and process video data, allowing it to assess whether current lifting feats are truly more impressive or simply amplified by algorithmic promotion. This research holds significance for the AI/ML community as it showcases practical applications of large language models (LLMs) and API integrations in analyzing social media trends. Utilizing an open-source vision-language model, the project implements innovative prompt engineering to extract and convert lifting data accurately. By applying structured responses and caching techniques, it optimizes the data extraction process, achieving an initial accuracy rate of about 73%. This study not only highlights the intersection of fitness culture and AI technology but also provides insights into creating efficient AI-driven data analysis frameworks, paving the way for further exploration in data-driven social phenomena.