The face of a criminal? A book about a controversial investigation that uses automatic facial recognition algorithms to predict whether or not a person will commit a crime is due to be published soon. However, more than 1,700 experts, researchers and academics in multiple fields of artificial intelligence have signed an open letter opposing such work , citing “serious concerns” about the study and urging Springer, the publisher that will publish the work, to withdraw its release.
A massive rejection
Using technology to predict crime would be “unfair and would cause real harm,” according to experts. The Coaltition for Critical Technology wrote an open letter to Springer Verlag in Germany expressing great concern about the newly developed automated facial recognition software that a group of scientists at the University of Harrisburg, Pennsylvania.
And it is that Springer’s Nature Research Book Series intends to publish an article by the Harrisburg scientists called “A deep neural network model to predict crime through image processing” (whose official statement about the study has been deleted from the network ). The coalition wants publication of the study, and others like it, to be rescinded, arguing that the paper makes claims that are based on shaky scientific premises, research and methods.
Scientists who have developed this AI stated that the algorithm is “80% accurate and without racial prejudice; the software can predict whether someone is a criminal based solely on an image of their face. The software is intended to help the police to prevent crime “.
Many experts have categorically denied its use, in this way:
“Let’s be clear: There is no way to develop a system that can predict or identify ‘criminality’ that is not racially biased, because the category of ‘criminality’ itself is racially biased,” he added. “Data generated by the criminal justice system cannot be used to” identify criminals or predict criminal behavior. Never”.
“Countless studies have shown that people of color are treated more harshly than similarly situated white people at all stages of the legal system, resulting in serious distortions in the data,” the expert group’s letter continues.
“Therefore, any software created within the existing criminal legal framework will inevitably echo those same fundamental biases and inaccuracies when it comes to determining whether a person has the ‘face of a criminal.'”
Springer, for its part, has stated that the document was not “at any time” accepted for publication and was submitted to a conference in which Springer will publish the minutes in which it had been rejected when the open letter was issued.
Meanwhile, organizers of the Coalition for Critical Technology have noted that ‘police science’ has been used in the past as a way to justify rationally discriminatory practices and have demanded that “all publishers refrain from publishing similar studies in the future”.
