Single-Turn Debate Does Not Help Humans Answer Hard Reading-Comprehension Questions

Alicia Parrish, Harsh Trivedi, Ethan Perez, Angelica Chen, Nikita Nangia, Jason Phang, Samuel Bowman


Abstract
Current QA systems can generate reasonable-sounding yet false answers without explanation or evidence for the generated answer, which is especially problematic when humans cannot readily check the model’s answers. This presents a challenge for building trust in machine learning systems. We take inspiration from real-world situations where difficult questions are answered by considering opposing sides (see Irving et al., 2018). For multiple-choice QA examples, we build a dataset of single arguments for both a correct and incorrect answer option in a debate-style set-up as an initial step in training models to produce explanations for two candidate answers. We use long contexts—humans familiar with the context write convincing explanations for pre-selected correct and incorrect answers, and we test if those explanations allow humans who have not read the full context to more accurately determine the correct answer. We do not find that explanations in our set-up improve human accuracy, but a baseline condition shows that providing human-selected text snippets does improve accuracy. We use these findings to suggest ways of improving the debate set up for future data collection efforts.
Anthology ID:
2022.lnls-1.3
Volume:
Proceedings of the First Workshop on Learning with Natural Language Supervision
Month:
May
Year:
2022
Address:
Dublin, Ireland
Editors:
Jacob Andreas, Karthik Narasimhan, Aida Nematzadeh
Venue:
LNLS
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
17–28
Language:
URL:
https://aclanthology.org/2022.lnls-1.3
DOI:
10.18653/v1/2022.lnls-1.3
Bibkey:
Cite (ACL):
Alicia Parrish, Harsh Trivedi, Ethan Perez, Angelica Chen, Nikita Nangia, Jason Phang, and Samuel Bowman. 2022. Single-Turn Debate Does Not Help Humans Answer Hard Reading-Comprehension Questions. In Proceedings of the First Workshop on Learning with Natural Language Supervision, pages 17–28, Dublin, Ireland. Association for Computational Linguistics.
Cite (Informal):
Single-Turn Debate Does Not Help Humans Answer Hard Reading-Comprehension Questions (Parrish et al., LNLS 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.lnls-1.3.pdf
Video:
 https://aclanthology.org/2022.lnls-1.3.mp4
Data
QuALITY