Biased Algorithmic Risk Assessment in criminal justice settings: How is COMPAS fraying the fabric of the right to a human decision-making in criminal procedure law

Artificial Intelligence-embedded technologies represent a blazingly new path, a foray into uncharted territory that holds a wealth, and a treasure trove, of scorching challenges that should not leave anyone lukewarm. When it comes to cutting-edge technology, you better not try your luck and the let the chips fall where they may. Artificial Intelligence-embedded technology, albeit a seemingly bedazzling gift to humankind, is not without multifarious caveats.

Amongst which stands a slew of perceived perils arising out of biased algorithm risk assessment in criminal justice settings that is shaping up to topple the foundations upon which stands algorithmic fairness. Hardly any bewildering surprise stems from the fact that Artificial Intelligence-embedded technology holds the blueprint of our destiny and the disruptive power it carries. This is about the unfolding of newer and higher forms of intrusiveness in our daily lives that extend beyond our collective grasp.

Building upon this, a couple of vexing questions crop up with bursting unease: should relevant stakeholders (i.e. judges, prosecutors, and lawyers) rely upon algorithmic risk assessment (i.e. COMPAS) in the purview of criminal justice? On the flip side, should machines be entrusted with the task of making high-stake decisions in the realm of criminal procedure law settings?

Bearing these burning (research) questions very firmly in mind, this paper adamantly contends that machines should not wield the power of making high-stake decisions that may bear on heavily citizens’ fundamental rights (i.e. the right to a due process, a central tenet of criminal procedure law) or inordinate harms may otherwise occur (i.e. the loss of functional reputation of the system of administration of justice).