Date of Award
2022
First Advisor
Kathryn Boswell
Second Advisor
Colette van Kerckvoorde
Third Advisor
Ansaf Salleb-Aoussi
Abstract
Among the contemporary issues facing the field of artificial intelligence (A.I.) is bias. Current implementations of artificial intelligence systems run on data created by individuals with implicit biases that can lead these systems to themselves becoming biased and contributing to the processes of structural violence, where institutions (e.g., education, healthcare, criminal justice) with which humans interact render the most marginal members of society even more vulnerable to these real yet largely invisible forces. History for this unconscious bias extends back to the field’s creation at Dartmouth University within an academic setting and continues today with industry’s use of A.I., with facial recognition as one timely concern. Anthropology focuses on people, cultural differences, and has historically tried to overcome human bias through its reflexive methods in which the anthropologist strives for self-awareness and transparency. This makes adopting an anthropological perspective a good fit for aiding A.I. in breaking the cycle of bias. By calling coders to be agents for change, anthropology encourages them to transform their beloved field by mitigating biases through reflexivity and to be conscious of the power they possess in creating systems with which humans daily interact. A two-part conversation, one focusing around the coders themselves and the other around the coding process, this thesis is the beginning of a larger conversation surrounding viewing artificial intelligence systems as contributors to human culture, filled with complexity and nuances. If artificial intelligence implements these changes, it will become even more of an aid to the human experience.
Recommended Citation
LoCascio, Cassidy, "Mitigating Bias in Artificial Intelligence through an Anthropological Lens: Introducing Humanity to A.I." (2022). Senior Theses. 1619.
https://digitalcommons.bard.edu/sr-theses/1619
Simon's Rock students and employees can log in from off-campus by clicking on the Off-campus Download button and entering their Simon's Rock username and password.
Comments
Ask at the Alumni Library circulation desk about the digital companion piece that accompanies this thesis.