The algorithm denied the benefit. No one decided. No one is responsible.
Literal meaning: Algorithmic violence describes harm produced by automated decision-making systems — benefits denials, misidentification by facial recognition, discriminatory credit scoring, wrongful criminal risk assessments — where the harm occurs through systemic bias or error in a system that no individual is directly responsible for in any given instance.
Origin: The concept was developed in critical algorithm studies and intersectional technology research, particularly by scholars including Safiya Umoja Noble (Algorithms of Oppression, 2018), Virginia Eubanks (Automating Inequality, 2018), and Joy Buolamwini (whose research on facial recognition bias produced the concept of “coded bias”). The term “algorithmic violence” crystallises a research tradition that documents how automated systems produce real, material harm — disproportionately to already-marginalised populations.
Harm produced by automated systems whose design encodes existing inequalities — where no individual decides, and no individual is accountable.
The Appeal: Automated decision systems offer efficiency, consistency, and the appearance of objectivity. For governments and organisations managing large populations — benefits claimants, credit applicants, criminal defendants — automation appears to eliminate the inconsistency and potential bias of individual human decisions. The scale and consistency of automated systems are real advantages.
The Friction: Informatics of Domination — Haraway’s 1985 framework — predicted this precisely: information systems reinscribe existing hierarchies while appearing neutral. Eubanks’ research documented that US benefits administration systems systematically disadvantaged poor families; Noble’s research documented that Google’s search algorithm produced racist results in searches for Black women. Dataism is the ideology that makes algorithmic violence invisible: if data is neutral, then data-driven decisions are neutral, and harm is the user’s problem rather than the system’s design. Surveillance Capitalism generates the data that encodes the existing inequalities that automated systems then reproduce at scale.
Why This Matters: Algorithmic violence names the specific accountability gap of automated harm. Once you see the structure — harm produced, no individual responsible — you can ask: who designed this system, what data trained it, whose interests does it serve, and who bears the cost of its errors?
Related terms: Informatics of Domination · Dataism · Surveillance Capitalism · Digital Exclusion · VSD (Value Sensitive Design) · Brussels Effect · Dehumanization · Vendor Lock-in
Read more:
- Automating Inequality — Eubanks, V. (2018). St. Martin’s Press
- Algorithms of Oppression — Noble, S.U. (2018). NYU Press