Steering Recommendations and Visualising Its Impact: Effects on Adolescents' Trust in E-Learning Platforms

Transparency builds trust: When adolescents could see how their control actions affected the recommendations they received, trust in the e-learning platform increased significantly compared to those without this visualization feature.

User control promotes healthy challenge-seeking: Giving students control over their mastery level encouraged them to "reasonably challenge themselves" rather than either avoiding difficulty or becoming overwhelmed, suggesting that control mechanisms support better learning behaviors.

Visualization explains algorithmic decisions: Showing the impact of steering actions increased perceived transparency, helping students understand why certain exercises were recommended to them—a finding with broader implications for making AI systems more explainable to users.

Adolescents can reflect on algorithms: When given control and visibility into how recommendations work, teenagers demonstrated the ability to think critically about the underlying recommendation algorithm, suggesting young users are capable of engaging meaningfully with AI systems when given appropriate tools.

E-learning platforms underutilize user control: Despite research showing the benefits of control mechanisms in recommender systems, most e-learning environments don't currently offer students ways to steer their recommendations, representing a gap between research evidence and real-world implementation.