Ninth International Conference on Communication & Media Studies

Abstract

A multi-stage study methodology was used to analyze collected data with a specific moral large-decisions survey (n= 26) in undergraduate (non-tech careers) students from Colombia. Initially, correlations were conducted among variables to identify relationships, but low correlations indicated limited explanatory power for trust in technology. In response, a machine learning approach was adopted, developing a neural network model capable of capturing complex relationships and learning patterns. Using a dataset with sociodemographic variables and trust-related responses, 581 parameters were tuned in the model to capture relationship complexities. During training, a loss function measured the disparity between model predictions and actual trust values, with optimization algorithms minimizing the loss function and tuning parameters. However, the initial model’s evaluation on a separate dataset showed 66.63% accuracy, suggesting unaccounted factors influenced trust. The second model achieved 90% accuracy for extreme cases and 75.4% to 79.71% for intermediate data. Results reveal key factors influencing trust in technology. Higher socioeconomic status correlated with greater trust due to increased purchasing power. Rules or norms regulating technology positively influence trust by providing a clear framework. Conversely, negative experiences decreased trust, emphasizing the importance of minimizing negatives and fostering positives. Recreational technology use had minimal impact on confidence, suggesting trust is influenced by other, more relevant factors.

Presenters

Juan J Giraldo Huertas
Associate Professor, Department of Developmental and Educational Psychology, Universidad de la Sabana, Cundinamarca, Colombia

José Amorocho
PhD Student, Psychology, Universidad de La Sabana, Colombia