Abstract
The current usage and application of ChatGPT and other AI systems demonstrate that human intelligence can be transcribed into machine learning and replicated algorithmically. This has raised various questions about machine learning, AI, and natural language processing (Anson and Straume, 2022; Graham, 2023; Johnson, 2023; Vee, 2023; MLA-CC Joint Taskforce on Writing and AI, 2023). Scholars in the field of technical and professional communication are increasingly interested in examining the AI and machine learning system, including the rhetoric on which it is built and the cultural fabric it will create.
This paper emphasizes the importance for scholars to scrutinize data sources and types, specifically focusing on training data and its relevance, in the development of AI reasoning devices such as Wearable Reasoners.It emphasizes that what is ingrained in the system holds rhetorical significance, and the process of incorporation, as well as the types of outputs generated, are equally crucial. The article presents findings based on the utilization of transrhetorical practices (Wang, 2021) in conjunction with a Data Feminist approach (D'Ignazio and Klein, 2020). The analysis focuses on the processed data within the IBM (International Business Machine) debater claim and evidence database, initially established in 2015 and yet to be updated. Additionally, it critically evaluates the implications and consequences of utilizing such data in the training of wearable devices, exemplified by the Wearable Reasoner created in 2020 using the IBM debater claim and evidence dataset from 2015.
As an individual embodying what Chandra Mohanty refers to as "Two Third World in One-Third World" or being "a part of the social minority now, with all its privileges," I inquire from the perspective of "a person situated in the One-Third World, but from the space and vision of, and in solidarity with, communities in struggle in the Two-Thirds World" (Mohanty, 1998, p. 507). Essentially, I am concerned about whether the data accurately represents gender, particularly in the context of third-world gender and politics. My transnational feminist positionality guides my examination of the homogenizing, universalizing, and consequently othering tendencies within data, data systems, and data architecture (Aguilar, 2022).