We study a fundamental problem in the evaluation of large language models that we call training on the test task. Unlike wrongful practices like training on the test data, leakage, or data contamination, training on the test task is not a malpractice. Rather, the term describes a growing set of techniques to include task-relevant data in the pretraining stage of a language model. We demonstrate that training on the test task confounds both relative model evaluations and claims about emergent capabilities. We argue that the seeming superiority of one model family over another may be explained by a different degree of training on the test task. To this end, we propose an effective method to adjust for training on the test task by fine-tuning each model under comparison on the same task-relevant data before evaluation. We then show that instances of emergent behavior largely vanish once we adjust for training on the test task. This also applies to reported instances of emergent behavior that cannot be explained by the choice of evaluation metric. Our work promotes a new perspective on the evaluation of large language models with broad implications for benchmarking and the study of emergent capabilities.

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 language models' survey responses on the basis of the well-established American Community Survey (ACS) by the U.S. Census Bureau. Using a de-facto standard multiple-choice prompting technique and evaluating 40 different language models, hundreds of thousands of times each on questions from the ACS, we systematically establish two dominant patterns. First, models have significant position and labeling biases, for example, towards survey responses labeled with the letter "A". Second, when adjusting for labeling biases through randomized answer ordering, models across the board trend towards uniformly random survey responses. In fact, binary classifiers can almost perfectly differentiate between models' responses to the ACS and the responses of the US census. Taken together, our findings suggest caution in treating survey responses from language models as equivalent to those of human populations at present time.