Name: Modelo matemático predictivo destinado a identificar perfiles criminales (Chile)
Area of Application: Policing
What kind of AI according to the project's official information: Predictive identification
Goals: Predict a person’s lifetime criminal career throughout life
How does it work according to the project's official information: According to Deputy Commissioner Angelo Dini, of CENACRIM: “We will no longer look for places where crimes are recurrent, now we will target perpetrators and those who may be prone to commit said crimes, narrowing down the search for possible investigation targets; this is because people and their behaviors are predictable according to their history.”
Some of the data gathered by the model are: prior arrests, prior convictions (antecedents), gun registry (DGMN), travel history, assets owned, education, family and/or criminal associations and ties.
Possible bias: Stigmatization, discrimination and criminalization of most vulnerable people.
Even researchers developing the model have warned about ethical dilemmas for deploying it.
Public Institutions Involved: Investigative Police of Chile (PDI) – National Center for Criminal Analysis (CENACRIM, Centro Nacional de Análisis Criminal)
Companies/Private Institutions Involved: Institute for Mathematical and Computational Engineering of the Pontifical Catholic University of Chile (IMC-UC, Instituto de Ingeniería Matemática y Computacional de la Universidad Católica)
Other interesting facts: Poses itself as the first experience in Latam (in July 2018 a prototype was already being tested)