ExplainabilityEthicsAccountabilitycs.HC
Grounding Explainability Within the Context of Global South in XAI
Deepa Singh, Michal Slupczynski, Ajit G. Pillai, Vinoth Pandian Sermuga Pandian
May 13, 2022
arXiv.org
3 citations
Key Takeaways
-
Explainability cannot be one-size-fits-all: The paper argues that current XAI (Explainable AI) research is largely developed in Global North contexts and may not adequately address the needs of Global South populations, suggesting that explainability standards must be tailored to specific social and cultural contexts rather than applied universally.
-
Social context matters for AI transparency: The authors emphasize that explainability should be "grounded" in the socio-technical systems and social contexts where AI actually operates, meaning explanations must account for local institutions, practices, and community needs rather than existing in isolation.
-
Global South perspectives are underrepresented in XAI research: The paper identifies a significant gap in explainability research focused on Global South contexts, particularly India, indicating that most current XAI frameworks may not reflect the priorities or constraints of populations in these regions.
-
Responsible AI requires localized approaches: By positioning explainability within specific geographic and social contexts, the paper suggests that truly responsible AI deployment requires understanding how different communities interact with and are affected by AI systems, not just technical explanations of how algorithms work.
Abstract
In this position paper, we propose building a broader and deeper understanding around Explainability in AI by 'grounding' it in social contexts, the socio-technical systems operate in. We situate our understanding of grounded explainability in the 'Global South' in general and India in particular and express the need for more research within the global south context when it comes to explainability and AI.