2026-09-21T01:15:54Zhttps://ecommons.cornell.edu/server/oai/requestoai:ecommons.cornell.edu:1813/1157572026-05-15T19:49:49Zcom_1813_35col_1813_47
Vecchione, Briana
Levy, Karen
Barocas, Solon
Kleinberg, Jon
2024-10-16T20:19:52Z
2024-10-16T20:19:52Z
2023-12
ProQuest Submission ID: 14034
ProQuest Publication ID: 30689365
https://hdl.handle.net/1813/115757
http://doi.org/10.7298/9826-gh79
16454776
243 pages
The field of algorithmic auditing is witnessing remarkable growth in response to concerns around discrimination-related harms exhibited by AI systems. However, this expansion has brought ambiguity regarding the definition of "audit" and the scope of evaluations. Although algorithmic audits involve various activities, defining an effective audit can be complicated due to normative, technical, and resource limitations. Challenges abound, from the lack of standardized language to diverse auditor characteristics. Nevertheless, the need for holistic evaluation is imperative as the societal impacts of algorithms become more apparent. Bridging theory and practice, this dissertation seeks to provide insights into and considerations for the algorithmic auditing community. By drawing on the history of audit studies in the social sciences, confronting the challenges of performing an audit in practice, and investigating the broader auditing ecosystem, it explores questions around the purpose of an audit, methodologies and best practices, available resources, community growth, and accountability. By doing so, it aims to improve the efficacy of AI audits and accountability for AI systems.
en
Attribution 4.0 International
https://creativecommons.org/licenses/by/4.0/
AI auditing
algorithmic auditing
Navigating the Complexities of Algorithmic Auditing: Challenges and Considerations
dissertation or thesis
https://newcatalog.library.cornell.edu/catalog/16454776
application/pdf
Information Science
Cornell University
Doctor of Philosophy
Ph. D., Information Science
https://hdl.handle.net/1813/59810.2
Publication
Cornell Theses and Dissertations
Vecchione, Briana
5893a6ea-7af3-41d7-abc6-04bcd26ab5df
open.access
open.access
<Publication xmlns="https://www.openaire.eu/cerif-profile/1.1/" id="046801af-ff0c-4fa0-a1d3-5a905b08e12f">
<Type xmlns="https://www.openaire.eu/cerif-profile/vocab/COAR_Publication_Types">http://purl.org/coar/resource_type/c_1843</Type>
<Language>en</Language>
<Title>Navigating the Complexities of Algorithmic Auditing: Challenges and Considerations</Title>
<PublishedIn>
<Publication>
</Publication>
</PublishedIn>
<PublicationDate>2023-12</PublicationDate>
<DOI>http://doi.org/10.7298/9826-gh79</DOI>
<Authors>
<Author>
<DisplayName>Vecchione, Briana</DisplayName>
<Affiliation>
<OrgUnit>
</OrgUnit>
</Affiliation>
</Author>
</Authors>
<Editors>
</Editors>
<Publishers>
<Publisher>
<OrgUnit />
</Publisher>
</Publishers>
<License>https://creativecommons.org/licenses/by/4.0/</License>
<Keyword>AI auditing</Keyword>
<Keyword>algorithmic auditing</Keyword>
<Abstract>The field of algorithmic auditing is witnessing remarkable growth in response to concerns around discrimination-related harms exhibited by AI systems. However, this expansion has brought ambiguity regarding the definition of "audit" and the scope of evaluations. Although algorithmic audits involve various activities, defining an effective audit can be complicated due to normative, technical, and resource limitations. Challenges abound, from the lack of standardized language to diverse auditor characteristics. Nevertheless, the need for holistic evaluation is imperative as the societal impacts of algorithms become more apparent. Bridging theory and practice, this dissertation seeks to provide insights into and considerations for the algorithmic auditing community. By drawing on the history of audit studies in the social sciences, confronting the challenges of performing an audit in practice, and investigating the broader auditing ecosystem, it explores questions around the purpose of an audit, methodologies and best practices, available resources, community growth, and accountability. By doing so, it aims to improve the efficacy of AI audits and accountability for AI systems.</Abstract>
<Access xmlns="http://purl.org/coar/access_right"
>
</Access>
</Publication>