Congratulations to Emily Gray '25 & Hailey Peter '25 on winning 2nd Place in the Behavioral Science Category at the 2023 Belton Molloy University Science Fair!
Their project Garbage In, Garbage Out is about a close analysis of racial and gender bias on AI classifications and outcomes.
The increasing use of AI that has been integrated into society has contributed to expediting common human capabilities. Prevalently seen amongst the recruitment process, AI resume screening has reduced the time and increased efficiency of analyzing applicants. As AI is coded and created by man, it has been widely credited with containing biases. Previous research suggests that this method of hiring prevents unconscious human bias and allows large applicant pools to be better analyzed. However recent research has pointed to socio-linguistic biases amongst responses which may further precipitate inaccurate description matching algorithms.
We have generated data from 48 fictional applicants, all of which obtain the same resume, but vary in ethnicity and gender. These individuals were assigned an ethnic, African-American derived, name or a generic, Caucasian, name. The resumes were run through an Open AI platform that generated varying response descriptions for candidates in an Amazon electrical engineering co-op position. A content analysis was performed, suggesting that similar to recent research, an underlying bias factor was exhibited between races. However, certain variables, such as the adjectives used were not significantly different across the subgroups. Our research has created an outlet for future research, and generated information to help fill the gap in present knowledge.