The Behavioral Design Podcast Recommender Systems with Carey Morewedge
Oct 23, 2024
Carey Morewedge, a Boston University marketing professor and expert in decision-making and AI systems, dives into the mechanics of AI recommender systems. He explains how platforms like Netflix and TikTok utilize user data to enhance engagement, often at the expense of long-term satisfaction. The conversation highlights ethical considerations, including the mismatch between stated and actual preferences, and the risks of over-relying on algorithms. Carey and the hosts discuss the delicate balance of human behavior and algorithmic influence, raising intriguing questions about technology's role in our lives.
AI Snips
Chapters
Transcript
Episode notes
Do But Don't Recommend
- Samuel Salzer plays a "do but don't recommend" game on Twitter.
- He avoids recommendations related to football, forcing himself to actively seek such content.
Self-Control over Algorithms
- Aline Holzwarth uses self-control to avoid engaging with content she doesn't want recommended.
- She actively resists clicking on videos she's tempted by, like parenting videos.
Recommender System Complexity
- Recommender systems are complex, using content-based filtering and collaborative filtering.
- Collaborative filtering uses implicit data like views and clicks, often prioritizing engagement over satisfaction.