Preliminary Conference Agenda

AI technologies, such as theme classification and named entity recognition, can enhance the accessibility and user-friendliness of digital library collections. However, they may introduce biases against marginalized groups if the collections behind the AI models do not represent them adequately. In the previous studies, AI models for poetry collections were developed without carefully assessing the datasets, raising concerns regarding representation. To address this issue, we annotated and published the race and ethnicity of poets in an American poetry collection curated by poets.org, which was recently used to train a poetry theme classification system. We then examined the diversity of the collection based on these annotations. Our findings indicate that most underrepresented groups are well-represented in the collection, which supports the dedication of the Academy of American Poets, the organization managing poets.org, to inclusivity and diversity. However, we found that poems by Latino/a/x American poets are less prevalent compared to their actual demographic representation. Furthermore, we found that poems from underrepresented groups increase the collection’s linguistic and thematic diversity, drawing on their unique cultures and histories. To design responsible AI that embraces diversity, it is important to support non-standard English and themes beyond those popular with the general population.