Author(s)

C. E. Cloete & J. S. Spies

Abstract

South Africa offers a valuable laboratory to study criminality because of its high levels of crime due to a heterogeneous environment of people, cultures, and economic development. This paper assesses the limitations of conventional practices in combating crime and illustrates the application of innovative artificial intelligence software to assumed unrelated databases – demographic, geographic, socio-economic and others – to find more effective ways to prevent criminal incidents. True unsupervised machine learning artificial intelligence software developed in South Africa provides neural network prediction of criminal incidents by identification of supposedly unrelated variables. Application of the software to find the hidden critical factors leading to bank robberies is illustrated in a specific case, where a robbery at a specific bank branch was predicted with an exceptionally high level of accuracy. Keywords: combating crime, artificial intelligence, causal layering analysis, neural networks, bank robberies, crime prediction.

Keywords

combating crime, artificial intelligence, causal layering analysis, neural networks, bank robberies, crime prediction