Investigating bias in technology through Baskin Engineering’s anti-racism research fellowship

roman-reggiardo.jpg

Roman Reggiardo, biomolecular engineering Ph.D. student

milad.jpg

Milad Hakimshafaei, computational media Ph.D. student

Technology is vulnerable to bias. Recognizing the need for anti-racist technology that takes into consideration the perceptions and experiences of underrepresented populations, the UC Santa Cruz Baskin School of Engineering established the Fellowship for Anti-Racism Research (FARR) in 2021.

Graduate students Roman Reggiardio and Milad Hakimshafaei are the latest fellows carrying out individual projects to research tools and techniques that will combat discrimination and racism in technology and engineering. 

“The idea for FARR emerged as we engaged in long, overdue discussions of how, as a school, Baskin Engineering can actively contribute to building a more just and anti-racist society,” said Alexander Wolf, dean of the Baskin School of Engineering. “FARR is one of a number of important school-wide initiatives that came out of those discussions, which also included the formation of the Baskin Inclusive Curriculum and Engineering Pedagogy (BICEP) initiative, active participation in designing the campus Faculty Equity Advocates program, and the appointment of Applied Mathematics Associate Professor Marcella Gomez as the school's first associate dean for diversity, equity, and inclusion.”

Eliminating discrimination for equitable medical technologies

Inherent bias has long been prevalent in genetic analysis technologies. As new medical technology is developed, it is imperative that adequate representation of diverse patient characteristics, including sex, race, and ethnicity, is accounted for early in the design process to ensure that products are effective, accessible, and equitable for all.

Machine learning tools play an increasingly important role in analyzing patient data, such as blood samples. These tools can inherit the bias of their technologists and can ultimately affect diagnoses. 

“We have tons of opportunities to accrue and encounter bias that may never be addressed,” said biomolecular engineering Ph.D. student Roman Reggiardo. “If we build a diagnostic tool that only works on one population, then we lose the opportunity to make the impact that we’re after. I wanted to investigate the current efforts to make diagnostics and liquid biopsies more equitable and how we can adjust things further to make it all unbiased moving forward.”

With mentorship from his faculty advisor Assistant Professor of Biomolecular Engineering Daniel Kim, Reggiardo is investigating the barriers to accessible and fair liquid biopsy diagnostics, a type of medical technology that uses cutting-edge DNA or RNA genetic sequencing to derive a complex diagnosis from a single blood draw. 

Reggiardo, who started his fellowship late summer 2022, is reviewing recent studies to assess patient populations and the potential biasing of early results from liquid biopsies. His aim is to both quantitatively and qualitatively assess the shortcomings of diagnostic performance and relevance across diverse populations. He then hopes to address these issues through new methodologies that draw from data representing a diverse population in order to develop technologies and computational tools that aid in fair and ethical practices.

“My thesis research is directly tied to how liquid biopsy technology can be equitably developed, deployed, and used,” Reggiardo said. 

Expanding diverse perceptions in visual design

Text-to-image artificial intelligence (AI) generators—systems that create images and patterns from written text descriptions—are becoming increasingly popular and are used for a wide range of content creation applications. 

Computational media Ph.D. student Milad Hakimshafaei is interested in using an AI system that generates visual images and patterns to collect information about perception and the aesthetic preferences of diverse populations to better understand and engineer more inclusive computational tools.

“As someone who came to the United States from the Middle East, I noticed the differences in lifestyle, access to technology, and opportunities to shape our future as humans,” Hakimshafaei said. “These experiences motivated me to get involved in research that highlights and addresses the areas of technology that need to be improved to be more inclusive.”

With mentorship from UCSC Open Source Program Office (OSPO) Incubator Fellow and PolyPhy Project Lead Oskar Elek and Stephanie Lieggi, executive director for OSPO and the Center for Research in Open Source Software, Hakimshafaei will develop an interface built off of an existing web-based AI generator that will allow the end user to control the system’s behavior to generate interesting and unique patterns. By studying a broad array of users’ visual design aesthetic preferences, he plans to develop algorithms that better represent the experience of beauty for different people. 

Through this fellowship, Hakimshafaei hopes to draw attention to the value of including a diverse representation of users in software development and the importance of open-source communities for sharing ideas and collaborating on projects to produce unbiased, anti-racist technology. 

Both Reggiardo and Hakimshafaei will present their research findings during the 25th anniversary of the Baskin School of Engineering, which will be celebrated throughout the 2022–23 academic year. FARR is open to all declared Baskin Engineering undergraduate and graduate students, including undergraduates pursuing engineering minors. Students may apply individually, or in teams of up to three students.

For more information on programs and resources available to students, staff, and faculty that support inclusive excellence at Baskin Engineering, please visit the Baskin Engineering Inclusive Excellence Hub.