Making Sense of Recommendations

Abstract

Studies how people interpret and use algorithmic recommendations for interpersonal predictions. The research shows that algorithms outperform human intuition at predicting others' preferences, yet people systematically prefer their own judgment.

Key Findings

• Algorithms outperform human intuition in predicting others' preferences
• People overweight their own social judgment relative to algorithmic predictions
• The gap between algorithm and human performance widens for less familiar others

Source