Algorithmic Determinism in the Justice System
Risk scores can influence decisions without determining a person's future.
The Mechanism: Predictive systems use historical data to estimate outcomes such as reoffending or missed court appearances. Their performance depends on data quality, definitions, and the way officials use scores; a risk estimate is not a deterministic forecast.
The Philosophical Pivot: Justice presumes that people are more than statistical resemblance to others. When a score changes someone's liberty, the institution must explain how uncertainty, unequal policing, and the possibility of change were handled.
The Horizon: Automated judgment can make old patterns appear objective because they arrive as numbers. A fair system needs contestability, evidence of impact, and human accountability that is more than a signature on a screen.
A score may change a person's access to liberty, housing, or opportunity long before they know how it was produced. Due process must therefore include a chance to inspect and challenge the basis of consequential decisions.
No metric can decide what society ought to value. That remains a political choice, and hiding it inside a model does not make the choice less human or less accountable.
A probability becomes a sentence when no one is allowed to challenge the number.
Would I accept being judged by a pattern drawn from people who were never truly like me?