Hannah Rose Kirk

A new test suite evaluates safety behaviors in language models.

A community-led initiative to identify harmful prompts in T2I models.

Research challenges traditional methods of evaluating language model values and opinions.

Examining community engagement and model adoption in open source AI development.

Small design changes can boost critical thinking when using language models.

Explore the need for an open feedback system to improve AI responses.

New methods improve language models’ understanding of user choices.