Finale Doshi-Velez & Been Kim (2017)
arXiv preprint.
URL: https://arxiv.org/abs/1702.08608
Abstract. Survey paper proposing a taxonomy of interpretability research — application-grounded vs human-grounded vs functionally-grounded evaluation — with explicit attention to which user populations benefit from each. The framing reference for XAI evaluation methodology before CHI took over the area.
Tags: ai-usability xai methodology