Artificial Intelligence and Emerging Technologies
How do AI and Emerging Technologies Affect Adolescents?
Since the rise of large language models and LLM-based chatbots like ChatGPT, AI has emerged as one of the foremost technologies reshaping our lives across fields like education, art and healthcare. A growing number of adolescents use AI chatbots for schoolwork, advice, companionship, and emotional support.
And this is only the beginning. AI will continue to transform our lives in ways we can barely anticipate, changing how we learn, work, love, relate to one another, and understand ourselves. Some of those changes may be beneficial; others may carry serious costs. Early research is already linking AI chatbot use to harmful outcomes including increased loneliness, reduced socialization, worsened academic performance, and even delusional spirals.
Figure 1: Regression plots showing the association between daily usage of each chatbot modality (text, neutral voice, and engaging voice) and psychosocial outcomes. Source: Fang et al. (2025). How AI and human behaviors shape psychosocial effects of chatbot use: A longitudinal randomized controlled study. MIT Media Lab.
This figure consists of four side-by-side regression plots that each show the association between daily duration of chatbot usage on the x-axis and a different psychosocial outcome (loneliness, socialization with people, emotional dependence on AI chatbots, and problematic use of AI chatbots) on the y-axis. Each plot shows three trend lines representing chatbot modality: text is represented by a light blue line, neutral voice by a medium blue line, and engaging voice by a dark blue line. The left-most plot shows that loneliness increased as daily duration increased. The slopes for the three modalities are similar, but steepest for the neutral voice modality. The second left-most plot shows that socialization with people decreased as daily duration increased. The slope is steepest for the neutral voice modality. The third left-most plot shows that emotional dependence on AI chatbots increased as daily duration increased. The slopes are similar, but steepest for the neutral voice modality. The right-most plot shows that problematic use of AI chatbots increased as daily duration increased. The slope is steepest for the neutral voice modality.
Figure 2: Plot showing the association between homework scores and exam scores for students who use AI and students who never used AI. Source: Does AI stop children from learning? (2026, August 18). The Economist.
This chart, titled “Thinking about it,” plots homework scores on the x-axis against exam scores on the y-axis. Both homework and exam scores are indexed so that the average score without AI use equals 100. The red line represents the association between homework and exam scores for students using AI, and the gray line represents the association between homework and exam scores for students who never used AI. Both lines follow a similar upward trend until the homework score reaches 100, with the gray line slightly above the red line. Past the point where homework scores equal 100, the gray line continues to trend upwards while the red line trends downwards. The downward trend becomes more steep as homework scores increase past around 115, showing an association between very high homework scores and worse exam scores for students using AI.
Our goal is to understand this transformation as it unfolds: how adolescents are using AI, how those uses affect their social, emotional, and cognitive development, which applications help or harm them, and how these effects change as the technology itself rapidly evolves.
Selected lab publications, essays, and resources
- Haidt, J., & Rausch, Z. (2025, Nov 24). Don’t give your child any AI companions. After Babel.