Systemic Biases in Sign Language AI Research: A Deaf-Led Call to Reevaluate Research Agendas

TL;DR

This paper critiques sign language AI research for systemic biases driven by hearing, non-signing leadership. Through a hybrid literature review of 101 papers, it reveals a predominant focus on mitigating perceived communication barriers, reliance on non-representative data, linguistically weak annotation schemes (notably glosses), and modeling choices that embed existing biases. It argues for Deaf leadership and greater transparency about researchers' positionalities to realign research with Deaf stakeholder needs and linguistic realities. The authors propose structural shifts toward inclusive governance, linguistic grounding, and accountable collaboration to ensure sign language technologies serve diverse Deaf communities without perpetuating inequities.

Abstract

Growing research in sign language recognition, generation, and translation AI has been accompanied by calls for ethical development of such technologies. While these works are crucial to helping individual researchers do better, there is a notable lack of discussion of systemic biases or analysis of rhetoric that shape the research questions and methods in the field, especially as it remains dominated by hearing non-signing researchers. Therefore, we conduct a systematic review of 101 recent papers in sign language AI. Our analysis identifies significant biases in the current state of sign language AI research, including an overfocus on addressing perceived communication barriers, a lack of use of representative datasets, use of annotations lacking linguistic foundations, and development of methods that build on flawed models. We take the position that the field lacks meaningful input from Deaf stakeholders, and is instead driven by what decisions are the most convenient or perceived as important to hearing researchers. We end with a call to action: the field must make space for Deaf researchers to lead the conversation in sign language AI.