Do LLMs exhibit human-like response biases? A case study in survey design

Lindia Tjuatja, Valerie Chen, Tongshuang Wu, Ameet Talwalkar, Graham Neubig

Primary Area: societal considerations including fairness, safety, privacy

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: large language models, evaluation, biases, computational social science

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

Abstract: As large language models (LLMs) become more capable, there is growing excitement about the possibility of using LLMs as proxies for humans in real-world tasks where subjective labels serve as the ground truth. A barrier to the adoption of LLMs as human proxies is their sensitivity to prompt wording. But interestingly, humans also suffer from issues of sensitivity to instruction changes. As such, it is necessary to investigate the extent to which LLMs also reflect human sensitivities, if at all. In this work, we use survey design as a case study, where human response biases caused by permutations in wordings of "prompts" have been extensively studied. Drawing from prior work in social psychology, we design a dataset and propose a framework to evaluate whether LLMs exhibit human-like response biases in survey questionnaires. Over the seven models we evaluated, we find that all but one (Llama2-70b), in particular instruction fine-tuned models, do not consistently display human-like response biases, and even sometimes show a significant change in the opposite expected direction of change in humans. Furthermore, even if a model shows a significant change in the same direction as humans, we find that perturbations that are not meant to elicit biased behavior may also result in a similar change, suggesting that such a result could be partially due to other spurious correlations. These results highlight the potential pitfalls of using LLMs to substitute humans in parts of the annotation pipeline, and further underscore the importance of finer-grained characterizations of model behavior.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 6211