Algorithmic, Epistemic, and AI Paternalism: Who Decides What You Get to Know
Recommender systems shape access to knowledge by optimizing for engagement rather than well-being, and AI increasingly decides on its own when to nudge or restrict the user.
The most inconspicuous form of contemporary paternalism governs not actions but access to knowledge itself.
The recommender as choice architect
Recommender algorithms act as choice architects at scale: they determine what information is available to the user at all, which sources are ranked higher, and which signals are treated as trustworthy (Communication Theory).
Epistemic paternalism
Epistemic paternalism is interference in access to information "for the good" of the subject without their consent. In the digital environment it becomes algorithmic and almost invisible: the person does not choose the filter; the filter chooses for them.
Engagement versus well-being
The key conflict lies in the metric. Optimising for engagement can diverge from the user's well-being: what holds attention is not always good for the person whose attention is being held (AlgEpi).
Epistemic welfare
In response, a normative framework of epistemic welfare has been proposed — the creation and maintenance of the conditions and capabilities for a person's epistemic agency in the digital public sphere (Communication Theory). Even "diversifying" recommenders that nudge toward variety reproduce the paternalistic logic of enlightenment.
Algorithmic paternalism
As AI spreads, "algorithmic paternalism" also emerges: the agentic appearance and attributed authority of systems can undermine a person's autonomy in decision-making (Frontiers in AI). The "smarter" a system seems, the more readily a person cedes the decision to it.
When AI decides to "nudge"
Applications for health, learning, and recommendation increasingly decide for themselves when to nudge, restrict, or heighten a user's attention. If a system optimises for engagement rather than well-being, the line between care and exploitation blurs. Hence the demands familiar throughout this site: transparency of purpose, the ability to opt out without penalty, and respect for a person's long-term interests.
Excerpts and dates
- 01к разделу «Эпистемическое благополучие»
Эпистемическое благополучие как рамка
«Epistemic welfare [is defined] as creating and maintaining the conditions and capabilities for individuals’ epistemic agency in the public sphere.»
Перевод: эпистемическое благополучие определяется как создание и поддержание условий и способностей для эпистемической агентности людей в публичной сфере.