Sociotechnical harms of algorithmic systems: Scoping a taxonomy for harm reduction
January 16, 2025
What are the risks from AI?
This week we spotlight the eleventh risk framework included in the AI Risk Repository:
Shelby, R., Rismani, S., Henne, K., Moon, A., Rostamzadeh, N., Nicholas, P., Yilla-Akbari, N., Gallegos, J., Smart, A., Garcia, E., & Virk, G. (2023, August). Sociotechnical harms of algorithmic systems: Scoping a taxonomy for harm reduction. Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society.
Using a scoping review and reflexive thematic analysis of computing research on harms (n = 172), this paper presents an applied taxonomy of sociotechnical harms from algorithmic systems.
The taxonomy consists of five major categories and 20 subcategories:
Representational Harms
Stereotyping social groups
Demeaning social groups
Erasing social groups
Alienating social groups
Denying people the opportunity to self-identify
Reifying essentialist social categories
Allocative Harms
Opportunity loss
Economic loss
Quality of Service Harms
Alienation
Increased labour
Service/benefit loss
Interpersonal Harms
Loss of agency
Tech-facilitated violence
Diminished health and well-being
Privacy violations
Social System Harms
Information harms
Cultural harms
Civic and political harms
Socio-economic harms
Environmental harms
Key features of the framework and associated paper:
Discusses and gives examples of each of the five major harm categories in detail, including their subcategories, and how technical components and social dynamics interact to produce them.
Frames harms in terms of their impacts across micro, meso, and macro levels of society.
Final taxonomy builds on and references existing taxonomies, classifications and terminology.
Disclaimer:
This summary highlights a paper included in the MIT AI Risk Repository. We did not author the paper and credit goes to Shelby, R., Rismani, S., Henne, K., Moon, A., Rostamzadeh, N., Nicholas, P., Yilla-Akbari, N., Gallegos, J., Smart, A., Garcia, E., & Virk, G. (2023). For the full details, please refer to the original publication: https://doi.org/10.1145/3600211.3604673.
