Book chapter
Friendship in the Age of Artificial Intelligence: Insights from Jonathan Haidt's Research on Wellbeing
Abstract
This chapter takes friendship as the everyday site where the costs and benefits of an AI-mediated social order are most visible. Drawing on Jonathan Haidt's diagnostic work on adolescent mental health, platform design, and the social architecture of the past fifteen years, it argues that the displacement of embodied, reciprocal friendship by algorithmically curated and chatbot-simulated relation is not an incidental side-effect of AI development but a constitutive feature of how present-generation systems are built. The argument reads Haidt's empirical work alongside the Aristotelian distinction between friendships of utility, pleasure, and virtue, and alongside the philosophical literature on relational goods. It closes with concrete design principles for technology that supports, rather than displaces, the practices through which friendship actually grows: reciprocity, sustained attention, conflict and repair, and shared participation in embodied communities.
A note on this piece
What is artificial intelligence doing to friendship? Not in the sci-fi sense — robot companions, digital lovers — but in the much quieter sense most of us already live with. The phone in your pocket. The algorithmic feed that ranks who you see. The notification badge pulling attention away from the person across the table.
This chapter argues that the social architecture of the past fifteen years has rewired friendship at the level of its preconditions. Drawing on Jonathan Haidt’s work on adolescent mental health and platform design, and on the Aristotelian account of virtue friendship through the philosophical concept of relational goods, it asks what survives — and what doesn’t — when the medium through which we relate to each other is engineered to optimize for engagement rather than care.
The argument is sympathetic to AI as a tool. It’s hostile to designs that simulate friendship while hollowing out the practices that make friendship possible: reciprocity, sustained attention, the work of repair after rupture. An AI companion can comfort. It cannot need you back. That asymmetry isn’t a minor design defect; it’s what makes the encounter something other than friendship.
The chapter sits in a tradition that takes seriously the claim that human persons are constitutively relational — that we become who we are inside a web of others whose flourishing is bound up with ours. Haidt’s empirical work gets most of the way to that claim from a secular angle; Aristotle and the personalist tradition fill in the rest.
A longer adaptation — Friendship in the Age of AI — is forthcoming as the inaugural essay on this site. The chapter is the argument; the essay will be its doorway for the general reader.