Investigating algorithmic bias in adaptive learning platforms

Ryan Baker, Jaclyn Ocumpaugh, and Nigel Bosch have been awarded $330,580 from the National Science Foundation to investigate algorithmic bias in adaptive learning platforms designed for 6th-12th grade math, using a variety of techniques to explore how students construct their own identities. Steve Ritter (CMU, co-PI) and Stafford Hood (UIUC) are also collaborators on this grant. The team will examine how bias may impact machine learned predictions of a wide range of constructs related to learning, and will use this information to conduct a large-scale experimental study about the students’ self-conceptions of identities to determine whether a variety of demographic factors is susceptible to algorithmic biases. They will then work to adapt these learning systems to ensure equitable opportunities for all learners.