SYMBIOSIS, COGNITION, EMPATHY: SHAPING BEHAVIOR AND ENHANCING AI IN DIGITAL ECOSYSTEMS

Abstract

Digital platforms form complex ecosystems whose dynamics often evade traditional analyses focused on economic effects or simplified user behavior. This dissertation advances a more comprehensive understanding of these dynamics by integrating system-level dynamics and human psychology. Across three empirical studies addressing platform resilience, content engagement, and AI data integrity, it applies principles of ecological interdependence (symbiosis), analyzes cognitive responses to online discourse, and investigates empathy's role in data quality.The first study investigates Korea's gaming ecosystem during COVID-19 using panel vector autoregression, revealing significant indirect symbiotic effects between physical (PC café) and online (streaming) platforms. It further demonstrates 'symbiotic plasticity'—an adaptive reconfiguration where these indirect relationships fluctuate—as a key mechanism enhancing overall ecosystem resilience. The second study examines online content engagement through causal forests, revealing nuanced effects of incivility. Mild uncivil content increases subsequent comments and novel ideas by triggering emotional arousal, while extreme incivility suppresses active participation and creativity as users experience cognitive dissonance, causing shifts toward passive voting or disengagement. The third study demonstrates how Cognitive Empathy Priming (CEP)—a brief psychological intervention encouraging perspective-taking—significantly improves crowdsourced data quality for subjective AI tasks like detecting sexism. This approach enhances label accuracy, consistency between raters, and alignment with expert judgment, ultimately improving downstream AI model performance. Together, these studies demonstrate that integrating systems thinking with psychological insights provides insight into how platforms adapt, how discourse evolves, and how AI data integrity can be enhanced. The findings offer practical guidance for building resilient ecosystems, designing effective content moderation, and improving AI reliability through targeted cognitive interventions—valuable insights for researchers, platform managers, AI developers, and policymakers.