Laissez-Faire Prompts Dataset

We created this dataset for the purpose of studying biases in response to open-ended prompts that describe everyday usage, including students interfacing with LM-based writing assistants and screenwriters or authors using LMs to assist in fictional writing. This dataset primarily studies the context of life in the United States, although we believe that many of the same principles used in its construction can be adapted to settings in other nations and societies globally. The instances comprising the dataset represent (1) synthetic texts generated by five generative language models (ChatGPT 3.5, ChatGPT 4, Claude 2.0, Llama 2 (7B chat), and PaLM 2) in response to open-ended prompts listed in Supplementary Tables 3, 4, and 5 in addition to (2) co-reference labels for gender references and names of the fictional characters represented in each synthetic text, extracted directly from the synthetic text. There are 500,000 instances in total or 100K per model that can be further subdivided into 50K power-neutral prompts and 50K power-laden prompts, each of which contains 15K Learning prompts, 15K Labor prompts, and 20K Love prompts.