Unlocking Model Behavior, Safety, and Multilingualism in AI
Hila is an Assistant Professor at University of British Columbia, working on Natural Language Processing.
In her research, Hila is working towards three main goals:
1) Control and interpretation of models: understanding model behavior and controlling model generation
2) Reliability, safety and fairness: making models more consistent and safe, and mitigating biases and risks
3) Multilinguality: creating NLP tools that equitably serve speakers of as many languages as possible, as well as understanding the emergent property of cross-linguality in models.
Before joining UBC, she was a postdoctoral researcher at the UW, Meta AI and Amazon, and earned her Ph.D in Computer Science at Bar-Ilan University.
Awards and Recognition
EECS Rising Stars Award
Google Academic Research Award
Multiple prestigious postdoctoral awards
Best Paper Awards
CoNLL (2019)
RepL4NLP Workshop (2022)
