📊 Planned Analysis
To move beyond simple "behavioral compliance," we will employ computational modeling to understand why the intervention works.
modeled via Mixed-Effects Logistic Regression
Pre/Post test difference (Interaction Effect)
Correlation between "Perceived Fairness" and "Latency"
The Hypothesis: High adoption rates paired with high autonomy scores will predict lasting behavioral change (Social Learning) rather than just suppression.
📚 Theoretical Underpinnings
This prototype explores AI-Mediated Communication (AI-MC). The goal is to find a balance between teaching youth towards positivity in social spaces while preserving user autonomy.
Research Questions: We are measuring acceptance rates and prosociality scores to determine if exposure to these interventions truly makes people more prosocial over time, or if it simply masks intent.
- Real-time toxicity scoring extensions like CLRSKY.ai for Bluesky.
- Research on the pros & cons of AI modifying human speech: Revisiting the Design of AI-Mediated Communication (2023).