Dissertations, Theses, and Capstone Projects

Date of Degree

2-2026

Document Type

Master's Thesis

Degree Name

Master of Arts

Program

Political Science

Advisor

Charles Tien

Subject Categories

American Politics | Political Science

Keywords

Artificial Intelligence, Political Redistricting, Minority Representation, Algorithmic Bias, Voting Rights Act, Gerrymandering

Abstract

Technology plays a significant role in people's daily lives worldwide. It has infiltrated many aspects of our lives, from education and healthcare to politics. With our increasing reliance on technology, artificial intelligence (AI) has become well-enmeshed in our political systems, affecting people everywhere, especially minority groups. This essay will investigate the potential effects of AI on the political influence of racial and ethnic minorities, specifically focusing on its role in altering voting district boundaries— a practice referred to as redistricting. There are several areas of concern under which AI could unintentionally worsen gerrymandering, impact equity in the representation of minority groups, and have technological implications regarding algorithmic bias and transparency, as well as fairness and representation regarding legal and policy decisions. The technological changes that arise with the use of AI could potentially make it much more difficult for minority groups to assert themself in politics, further marginalizing their voices. By examining and evaluating this issue, I hope to provide insight into how AI-driven redistricting algorithms may perpetuate gerrymandering practices that diminish the political power of racial and ethnic minority communities, further marginalizing their voices in the political process, bringing into question the fairness and inclusivity of our democratic system.

The use of AI algorithms in redrawing electoral district boundaries may have both intentional and unintentional consequences. The optimization of specific criteria, such as equality, compactness, or partisan advantage, may result in biased outcomes that result in gerrymandered districting, which fragments racial and ethnic minority communities by separating them into different districts or concentrating them in a single district as part of a packing and cracking strategy. All of this may reduce minority communities' ability to elect representatives that truly represent their interests. How does algorithmic bias in AI-driven redistricting impact minority political representation compared to human-led redistricting?

AI redistricting and human-led redistricting have similar intentions: to create fair, balanced, and representative electoral districts. However, they differ significantly in their methodologies and potential for bias. Both approaches must maintain compliance with legal standards (i.e., population equality, compactness, and respect for community boundaries), and both encounter challenges such as preventing gerrymandering and guaranteeing effective representation of diverse communities. Human redistricting typically involves manual processes involving political intuition, judgment, and sometimes partisan interests. These manual processes often lead to subjective decisions, as humans may unintentionally or intentionally introduce biases based on personal or political motivations. This tendency can result in districts that favor one party or demographic over another, exacerbating existing inequalities.

AI redistricting, by contrast, utilizes data-driven algorithms to optimize district boundaries based on predefined criteria. The use of AI in the political redistricting process allows for greater efficiency and scalability than human-led redistricting, as it can process extensive amounts of demographic and geographic data at rates much faster than people. However, this does not mean that AI is empty of bias, as algorithms can reflect the biases in the data that is used to develop machine learning systems. This type of bias can lead to outcomes that may disadvantage certain groups. This is why transparency and accountability are pivotal in both approaches. While human redistricting processes are often more open to public scrutiny, AI processes require detailed documentation and open algorithms to address the "black box" problem, where the decision-making process of the AI is not easily understood by humans. Moreover, AI redistricting benefits from greater computational power and the ability to run numerous simulations to find the most balanced district configurations. However, human oversight remains essential. People are needed to design the algorithms, select and pre-process the data, interpret the results, and guarantee that the results comply with established legal and ethical standards. This oversight helps to mitigate biases in the process and adjust for any unintended consequences the AI might produce.

While AI can significantly enhance the redistricting process by handling large datasets and complex calculations more efficiently than humans, it requires continuous human involvement to ensure fairness and representation. Both AI and human redistricting aim for equitable outcomes, but the integration of AI offers a promising tool to support and improve the transparency and objectivity of the redistricting process, provided that it is carefully managed and monitored by humans. This thesis proposes that AI technologies, while potentially promising, may inadvertently perpetuate gerrymandering practices that marginalize minority voices in the political process.

Download

Share

COinS