Fetching the paper…

Proxy Tasks and Subjective Measures Can Be Misleading in Evaluating Explainable AI Systems · Around