Language models assign responsibility based on actual rather than counterfactual contributions.

Yang Xiang, Eric J. Bigelow, Tobias Gerstenberg, Tomer D. Ullman, Samuel Gershman

Browse the full CogSci paper archive.