Questioning the Survey Responses of Large Language Models

As large language models increase in capability, researchers have started to conduct surveys of all kinds on these models with varying scientific motivations. In this work, we examine what we can learn from a model's survey responses on the basis of the well-established American Community Survey (ACS) by the U.S. Census Bureau. Evaluating more than a dozen different models, varying in size from a few hundred million to ten billion parameters, hundreds of thousands of times each on questions from the ACS, we systematically establish two dominant patterns. First, smaller models have a significant position and labeling bias, for example, towards survey responses labeled with the letter "A". This A-bias diminishes, albeit slowly, as model size increases. Second, when adjusting for this labeling bias through randomized answer ordering, models still do not trend toward US population statistics or those of any cognizable population. Rather, models across the board trend toward uniformly random aggregate statistics over survey responses. This pattern is robust to various different ways of prompting the model, including what is the de-facto standard. Our findings demonstrate that aggregate statistics of a language model's survey responses lack the signals found in human populations. This absence of statistical signal cautions about the use of survey responses from large language models at present time.

On the Adversarial Robustness of Causal Algorithmic Recourse

Algorithmic recourse seeks to provide actionable recommendations for individuals to overcome unfavorable outcomes made by automated decision-making systems. Recourse recommendations should ideally be robust to reasonably small uncertainty in the features of the individual seeking recourse. In this work, we formulate the adversarially robust recourse problem and show that recourse methods offering minimally costly recourse fail to be robust. We then present methods for generating adversarially robust recourse in the linear and in the differentiable case. To ensure that recourse is robust, individuals are asked to make more effort than they would have otherwise had to. In order to shift part of the burden of robustness from the decision-subject to the decision-maker, we propose a model regularizer that encourages the additional cost of seeking robust recourse to be low. We show that classifiers trained with our proposed model regularizer, which penalizes relying on unactionable features for prediction, offer potentially less effortful recourse.