Where: ZOOM Join from PC, Mac, iOS or Android: https://vuw.zoom.us/j/93609230659

When: Wellington 9:30 AM – 10:30 AM Friday 05/08/2022

The quest to open black box Artificial Intelligence (AI) systems brought about the rise of the research field Explainable AI (XAI). XAI methods automatically generate explanations along with AI decisions. Researchers and practitioners alike underpin that XAI methods should be designed user-centric, i.e. address the needs of their users. Against this background, counterfactual explanations appear particularly promising, as they are in line with how humans construct explanations themselves.

The goal of this talk is to give insight into user-centric XAI and potential application scenarios. Specifically, the design and evaluation of an approach to produce coherent counterfactual explanations are presented and discussed.