Machine learning and forensic risk assessment: new frontiers

ABSTRACT

Advanced approaches to predicting offending are increasingly transpiring without the forensic psychology discipline’s involvement – an area that it has piloted and influenced for many decades. Computer science experts have built an impressive decade-long literature base on risk assessment – a technical literature that is not only progressing at a fast pace, but appears to function independently, for the most part, from the forensic risk assessment literature. This paper outlines the potential utility of machine learning approaches and the broader ‘algorithmic culture’, for forensic risk assessment, and the implications their use (and non-use) may have for the discipline.

KEYWORDS:

• Machine learning
• risk assessment
• forensic psychology
• offending

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Various methods are currently being developed and utilised to ‘open the black box’ and provide information about the relationships between predictor variables and outcomes (Berk, 2012; Lipton, 2017).

2. Some overlay has occurred in a related field (Criminology, see Berk, 2012; Berk et al., 2017).