LACES: Latin-American Community-Sourced Evaluation of Stereotypes

LACES addresses the critical geo-cultural gap in Natural Language Processing bias evaluations, where Latin American varieties and communities are historically under-represented.

• Publication: Findings of ACL 2026 (“Adaptive Data Collection for Latin-American Community-sourced Evaluation of Stereotypes”).
• Methodology: Implements an adaptive data collection protocol that iteratively expands and validates demographic stereotypes through community input.