What Is the Philosophy of Algorithms?

What Is the Subject?

An algorithm is, technically, a finite procedure for solving a problem. The philosophy of algorithms begins where the definition leaves off: what happens when these procedures govern human life. A hiring algorithm screens resumes, a credit algorithm sets interest rates, a social media algorithm curates reality, a policing algorithm allocates attention, and a sentencing algorithm influences liberty. None of these is a neutral calculation. Each encodes assumptions, reflects data, distributes benefits and harms, and answers to someone. The philosophy of algorithms studies what these systems are doing to fairness, transparency, power, and accountability.

The central discovery of the field is that algorithms are not mathematics plus neutrality. They are mathematics plus choices: choices about what to optimize, what data to use, what counts as success, and who gets to make those choices. The philosophy of algorithms is the study of those choices.

Historical Background

The formal idea of the algorithm is ancient, from Euclid's procedures to the medieval algorists, but the philosophical stakes changed when algorithms left the textbook and entered the institution. Credit scoring arrived in the 1950s and 60s, quietly shifting decisions from bankers to formulas. The internet and machine learning changed the scale: by the 2000s, algorithms were curating information, and by the 2010s, they were deciding on loans, hiring, and criminal justice.

The scandals did the philosophical work. ProPublica's 2016 investigation of the COMPAS recidivism algorithm showed that the same tool could be criticized as biased and defended as accurate, depending on the definition of fairness, and the result was a genuine philosophical discovery: fairness is mathematically underdetermined, and different reasonable fairness criteria contradict each other. Cathy O'Neil's Weapons of Math Destruction (2016) argued that harmful algorithms share a pattern: opacity, scale, and damage. The field has been developing the ethics and philosophy of these systems ever since.

The most important philosophical discovery of the field is that fairness is a choice, not a number. A lending model can be made to satisfy any one definition of fairness, equal treatment, equal error, equal outcome, and each definition fails one of the others. The mathematics does not decide which definition is right; the politics does, and the politics is usually hidden inside the choice of metric. The philosophy of algorithms insists that the choice be made visible and public, because the choice decides who the system protects and who it sacrifices, and the people who make it should have to answer for it.

Key Concepts

The first concept is algorithmic fairness and its paradoxes. Fairness can mean equal error rates across groups, equal outcomes, or equal treatment of equals, and these cannot all be satisfied simultaneously. The COMPAS debate proved it: a tool that equalizes false positive rates cannot also equalize calibration, and vice versa. The philosophy of fairness in algorithms is the study of these trade-offs and the question of who should choose among them.

The second concept is opacity and the black box. Some algorithms are secret by corporate design, some are opaque because of their complexity, and some are opaque because even their builders cannot fully explain them. Opacity conflicts with fairness and accountability: you cannot contest a decision you cannot understand, and you cannot assign responsibility for a process that is invisible. The demand for explainability is a philosophical demand, not just a technical one.

The third concept is the distribution of power. Algorithms are not just decision tools; they are instruments of power. Whoever controls the algorithm controls the decision, the data, and the criteria. The philosophy of algorithms asks how this power is legitimated and limited, and it notes the asymmetry: the system sees the citizen in enormous detail, while the citizen sees the system hardly at all. This asymmetry is a problem of justice, not just of design.

Contemporary Relevance

The relevance grows daily. Generative AI models are algorithms that produce text, images, and decisions, and they inherit all the questions: trained on the internet's biases, opaque in their reasoning, and deployed with enormous consequences and little accountability. The debates about AI regulation, about audits, transparency, impact assessments, and rights of explanation, are the philosophy of algorithms becoming law.

The practical conclusion is that we should treat algorithms as institutions, not as gadgets. Institutions get governed: they have oversight, appeals, audits, and accountability. Algorithms that make consequential decisions should get the same, and the philosophy of algorithms is the discipline that insists on it, asking not just what the algorithm does, but what it should do, and who should decide.

The second discovery is about power and its invisibility. The algorithm that decides your credit, your screening, your feed, is exercising power over you, and the power is harder to see and harder to contest than the power of a person, because there is no one to face, no reasons to demand, no appeal to make. The philosophers of power have always known that the strongest power is the power that disappears into the structure, and the algorithm is the purest form of structural power yet built. The response is accountability: the right to know, the right to contest, and the right to a human who answers.

Sources

• Stanford Encyclopedia of Philosophy, "Ethics of Artificial Intelligence and Robotics" — https://plato.stanford.edu/entries/ethics-ai/
• Stanford Encyclopedia of Philosophy, "Philosophy of Computer Science" — https://plato.stanford.edu/entries/computer-science/
• Cathy O'Neil, Weapons of Math Destruction (Crown) — https://www.penguinrandomhouse.com/books/241637/weapons-of-math-destruction-by-cathy-oneil/

Related Topics

• Machine Ethics — the moral design of automated systems.
• Information Ethics — values for data and information systems.
• Cathy O'Neil — the critic who exposed harmful algorithms.
• The Ethics of Social Media — algorithms curating human life.