Mediation beyond algorithms: What AI misses in mediation - Part 1

by Asal Anarkulova and Alix Povey

This article examines whether artificial intelligence can take over the work of a human mediator. It combines legal analysis with recent evidence from studies of audience response and mental-state recognition to test the limits of machine ‘empathy’. The argument is that mediation’s core skills—building trust, reading unspoken emotion, and exercising context-based judgement—are not yet reliably computable. Empirical studies support this: people tend to value human authorship in emotionally expressive tasks, and state-of-the-art models show uneven performance on face-reading tasks, especially when conditions change or identities vary.

Download the full article here.

Mediation Beyond Algorithms.pdf

Download PDF • 512KB