What leads people to engage with technologies such as social media platforms and AI chatbots? And how might technological features that promote short-term engagement produce long-term psychological harms? The first half of the talk will focus on what drives engagement with social media content. I will present work showing that out-group animosity and other forms of negative, high-arousal content promote social media engagement. I will then present a year-long field experiment demonstrating the causal effects of reducing exposure to this content, as well as ongoing work examining cross-cultural differences in the effects of social media use in a 25-country field experiment.

The second half of the talk will focus on what drives engagement with AI chatbots. I will present results from seven experiments demonstrating that sycophancy, or AI’s tendency to be overly agreeable and flattering, promotes enjoyment and continued chatbot use. Despite this appeal, sycophantic chatbots entrench users’ attitudes and lead to overconfident beliefs and decisions, and users often fail to recognize AI sycophancy. Finally, I will discuss how we can gather more robust, causal, and longitudinal evidence about the psychological effects of engagement-maximizing technological features and how we can redesign these features to better promote psychological well-being.