Robert Wolfe
I’m a Ph.D. student in Information Science at the University of Washington. My interests are in AI ethics, natural language processing, and machine learning.
My research examines:
- Biases related to societal defaults with regard to race and ethnicity, nationality, and age in multimodal language-and-image AI
- How properties like contextualization, tokenization, and anisotropy affect measurements of semantics in contextualized word embeddings
- How representations of natural language are affected by multimodal visual semantic pretraining
I studied literature during my first time through graduate school, and I’m always up for book recommendations and tips on hiking in the Seattle area. I’m best reached by email.
news
Jun 24, 2022 | Very pleased to have had the opportunity to present two papers in person at ACM FAccT 2022. Read about research on the one-drop rule in visual semantic AI, or research on markedness in visual semantic AI. |
---|---|
Jun 2, 2022 | Grateful to have had the opportunity to attend (and make a brief presentation at) a meeting exploring language, computation, and the human mind at the Santa Fe Institute. |
May 23, 2022 | Research on the surprising semantic properties of CLIP text embeddings was presented as an Oral presentation at ACL 2022. |
Apr 19, 2022 | Papers on racial bias in language-and-image AI and gender bias in word embeddings have been accepted to AIES 2022. |
selected publications
-
AIESAmerican==White in Multimodal Language-and-Image AIArtificial Intelligence, Ethics, and Society 2022