Abstract: Recent years have seen an increase in the use of social me-dia for various decision-making purposes in the context ofurban computing and smart cities, including management ofpublic parks. However, as such decision-making tasks arebecoming more autonomous, a critical concern that arises isthe extent to which such analysis are fair and inclusive. Inthis article, we examine the biases that exist in social media analysis pipelines that focus on researching recreationalvisits to urban parks. More precisely, we demonstrate thepotential biases that exist in different data sources for esti-mating the number and demographics of visitors through acomparison of image content shared on Instagram and Flickrfrom 10 urban parks in Seattle, Washington. We draw a com-parison against a traditional intercept survey of park visitorsand a multi-modal city-wide survey of residents. We eval-uate the viability of using more complex AI facial recognition algorithms and its capabilities for removing some ofthe presented biases. We evaluate the AI algorithm throughthe lens of algorithmic fairness and its impact on sensitivedemographic groups. We show that despite the promisingresults, there are new sets of concerns regarding equity thatarise when we use AI algorithms.