Concept bottleneck models map from raw inputs to concepts, and then from concepts to targets. Such models aim to incorporate pre-specified, high-level concepts into the learning procedure, and have been motivated to meet three desiderata: interpretability, predictability, and intervenability. However, we find that concept bottleneck models struggle to meet these goals. Using post hoc interpretability methods, we demonstrate that concepts do not correspond to anything semantically meaningful in input space, thus calling into question the usefulness of concept bottleneck models in their current form.
@article{arxiv.2105.04289,
title = {Do Concept Bottleneck Models Learn as Intended?},
author = {Andrei Margeloiu and Matthew Ashman and Umang Bhatt and Yanzhi Chen and Mateja Jamnik and Adrian Weller},
journal= {arXiv preprint arXiv:2105.04289},
year = {2021}
}