Agentic AI for Social Good Shows Geographic Bias in Research
A study analyzing 112 publications on agentic AI aimed at social good, released between 2015 and 2026, uncovers a moral-geographic imbalance. It was observed that 73% of the papers (82 out of 112) lack any geographic context. Research focusing on health or physical/ecological Sustainable Development Goals (SDGs) includes geographic details 37-40% of the time, whereas those related to institutional and social-policy SDGs do so merely 13% of the time. Notably, SDG 16, which addresses peace, justice, and strong institutions, is both the most frequently discussed goal and has the least geographic specification. The findings suggest that assertions of social good do not guarantee accountability to the communities that these systems aim to benefit.
Key facts
- Survey of 112 papers on agentic AI for social good
- Published between 2015 and 2026
- 73% of papers specify no geographic context
- Health or physical/ecological SDGs specify geography 37-40% of the time
- Institutional and social-policy SDGs specify geography only 13%
- SDG 16 is the most-covered goal and has the lowest geographic specification
- Claims of social good do not establish accountability to communities
- Moral-geographic asymmetry identified
Entities
Institutions
- United Nations
- arXiv