CITP Seminar: Beyond Algorithmic Bias: An Interrogation of the Google Search by Image Algorithm

We perform a socio-computational interrogation of the google search by image algorithm, a main component of the google search engine. By drawing from Bourdieu’s theory of cultural reproduction, we treat the algorithm as a subject, that was shaped (trained) by a specific culture and hence produces and reproduces the conditions of its creation. In this way, the algorithm functions as a window for understanding the attitudes of its designers, owners, and the ones existing in the dataset it was trained on.

We learn from the algorithm by presenting it with more than 40 thousands faces of all ages and more than four races (prompts) and collecting and analyzing the assigned labels (responses) with the appropriate statistical tools. In this way, we uncover a small portion of the algorithm’s beliefs, values, knowledge and skills, and use its visual understanding of humans and their appearance to learn the algorithm’s reflection of society. We find that the algorithm reproduces structures existing in the white male patriarchy, often simplifying, stereotyping and discriminating females and non-white individuals, while providing more diverse and positive descriptions of white men.

Since we are able to locate structural elements of the algorithm’s culture, we study up: we provoke individuals at the top of the techno-hierarchical ladder, by demonstrating how the algorithm places them within the socio-cultural reality that they consciously or unconsciously shaped, many times creating biased representations of them. Based on the analysis, we discuss the scientific and design implications of the study and provide suggestions for alternative ways to design just socioalgorithmic systems.

Article co-author: Arwa Michelle Mboya (MIT Media Lab)

Bio:
Orestis Papakyriakopoulos is a postdoctoral research associate at CITP. His research showcases political issues and provides ideas, frameworks, and practical solutions towards just, inclusive and participatory socio-algorithmic ecosystems through the application of data-intensive algorithms and social theories.

Orestis has a master’s degree in civil engineering from the National Technical University of Athens and a master’s degree in philosophy of science and technology from the Technical University of Munich. He received a Ph.D. in computer science from the Technical University of Munich. In 2019-20 he was a visiting scholar at the Massachusetts Institute of Technology’s Center for Civic Media.

To request accommodations for a disability please contact Jean Butcher, butcher@princeton.edu, at least one week prior to the event.