Predict, monitor, and punish: Algorithms forecasting crime.

Luis Avena, Postphenomenologist of Time. Mexico City

I was reading an article about “the University of Chicago having developed a new algorithm that forecasts crime by learning patterns in time and geographic locations from public data on violent and property crimes”. I found it a very interesting topic for several reasons. First, the issue of prediction from historical data and the application of algorithms reveals itself as a controversial practice in ethical and political terms. Second, the classification of crime is reconfigured from actions that are not being deliberate by a human mind. And third, it is a surveillance technology that anticipates up to a week before the criminal act. Why design technology to monitor and punish, and not to prevent and intervene?

The title of this short article could be a suggestion for the new version of Michel Foucault’s book Surveiller et Punir in which he gives a genealogy of the birth of social readaptation institutions, known simply as prisons. Among many and varied themes, the French thinker thematizes the disciplining of the bodies of individuals as a form of permanent surveillance based on the metaphor of the panopticon. In this disciplinary social model, control is and remains a fundamental action of institutions and government. However, today we find a factor that changes our understanding of the classification of crime in the judicial system: the time in which the criminal action is executed. Having the ability to predict a criminal action in the near future is an updated model of power technologies that monitor and punish actions and individuals.

Although it may not seem like it, time is and remains a relevant resource for the invention of new technology. From the little I have worked on this problem, I could say that the temporal relationship we will experience is one of compression of the classical categories of past, present, and future. That is, the past will be increasingly influencing our present in the form of data to make decisions. The future will be less and less uncertain and random because from a present point in time we will be able to predict and forecast from patterns all kinds of behaviors of objects and living beings. In a manner of speaking, time control is a fundamental factor of all coming technology. They start predicting crimes a week in advance! The truth is that I cannot imagine a world in which the behavior of any individual can be predicted by taking into account past data such as origin, friendships, parents’ profession, places he/she frequents, and people he/she talks to, websites he/she visits. All this can be related to criminal profiling and crime prediction.

How should this tool be used? The same article posits that there are biases in the algorithms in that police response is effective in urbanized spaces with good socioeconomic status, which makes groups that “historically” have committed crimes vulnerable. What is surprising is that we all know that the police are more effective in wealthy neighborhoods and negligent in those with fewer resources. In other words, algorithms work with the bias that “who has already stolen once, will surely do it again”. This is an interesting issue for the ethics and politics of algorithms. It seems clear to me that if technology is capable of 90% accurate prediction of “crimes”, then couldn’t this tool be used to prevent criminal action? That is, it seems questionable to me that the tool would be used to monitor and, once the crime is committed, to punish. This may be a clear example of how technology is designed to legitimize government institutions. That is, predicting crime with 90% efficiency does nothing but justify the penal system, which is primarily responsible for prosecution and punishment. My problem with this is that if it is possible to predict criminal acts, wouldn’t it be better to invent new institutions that approach people before they commit a crime in order to provide support to those who need it?

--

--

Luis Avena
Institute for Ethics and Emerging Technologies

Philosopher interested in Post-Phenomenology, AI, Emerging Technologies, Hedonistic Materialism, Camus’ Existentialism, Clinique Sociology and Anarchism.