Investigating Gender Bias in Artificial Intelligence: The case of GPT

<a href="https://embed.stanford.edu/iframe/?url=https%3A%2F%2Fpurl.stanford.edu%2Fkg187pj4784" class="su-underline">Show Content</a>

Abstract/Contents

Abstract: In the past decade, there has been rapid growth in the field of artificial intelligence (AI), with 2022 alone witnessing an investment of $91.9 billion in the private sector. This technological explosion, fueled by the advent of big data, computational power, and innovative methodologies, has not only amplified the efficiency, safety, and quality of work but also heralded an era of unprecedented innovation and economic progress. At the same time, the application of AI to complex problems in real-world situations raises many ethical concerns for policy makers and business leaders, including the presence of bias in AI systems and their consequent perpetuation thereof. This paper explores whether one of the most popular AI algorithms, OpenAI’s GPT, exhibits gender bias. The research compares GPT’s characterization of the typical gender of certain occupations with Bureau of Labor Statistics data on the actual gender composition of those occupations. The analysis for this paper showed that GPT did exhibit gender bias in 38 of the 50 occupations studied. This finding suggests that both researchers and policy officials might wish to give further attention to gender bias in AI systems and consider steps to mitigate that bias.

Type of resource	text
Publication date	August 18, 2023; June 23, 2023

Author	Sadeghi, Tina
Advisor	Windham, Patrick
Degree granting institution	Stanford University
Department	Public Policy Program

Subject	artificial intelligence
Subject	bias
Subject	algorithmic gender bias
Subject	OpenAI's GPT
Subject	gender bias
Subject	gender bias in artificial intelligence
Subject	Stanford University
Subject	Humanities and Sciences
Subject	Public Policy Program
Genre	Text
Genre	Thesis

DOI	https://doi.org/10.25740/kg187pj4784
Location	https://purl.stanford.edu/kg187pj4784

Use and reproduction: User agrees that, where applicable, content will not be used to identify or to otherwise infringe the privacy or confidentiality rights of individuals. Content distributed via the Stanford Digital Repository may be subject to additional license and use restrictions applied by the depositor.
License: This work is licensed under a Creative Commons Attribution 4.0 International license (CC BY).

Preferred citation: Sadeghi, T. (2023). Investigating Gender Bias in Artificial Intelligence: The case of GPT. Stanford Digital Repository. Available at https://purl.stanford.edu/kg187pj4784. https://doi.org/10.25740/kg187pj4784.

Stanford University, Public Policy Program, Masters Theses and Practicum Projects

Loading usage metrics...