The idea that a person’s character can be judged from their face has been around for thousands of years. The ancient Greeks believed in this pseudo-science, as did the early peoples of Egypt and India. In China during the Spring and Autumn period (770 – 481 BCE) historical texts record that "face-reading" – mianxiang – was used to determine character and honesty. Today in the western world the practice is known as physiognomy and there are still those who believe that some criminals inherit their criminality; such people are said to be "born-criminals."
In 19th-century Italy a dubious scientist called Cesare Lombroso invented the word "criminology" and the belief that criminals were born lawless and inherited villainous faces. Criminals, he said, could be identified by measuring their bones and heads.
This brings us to the present and the use of artificial intelligence to create a scientific version of physiognomy. The UK prison system is using an "artificial intelligence violence predictor" to assess prisoners’ crimes, age and character so as to predict which prisoners are the most likely to be dangerously violent.
This is just one example of a new technology called AI-profiling that is being used by more and more law enforcement agencies. It has been given the names "pre-crime technology" and "predictive policing" and is, in my admittedly non-scientific opinion, a weird and maybe scary kind of science. It envisages a future where the police arrest innocent people because pre-crime technology has assessed they are about to commit crimes. Police officers would have to say: “You are under arrest because it is believed you are about to commit a crime.” Carry the logic further and it is possible to imagine prisons that contain not old-fashioned criminals who have committed crimes but pre-criminals who have done nothing wrong.
Up to now such a world has been confined to science fiction books and sci-fi movies by directors like Steven Spielberg. Such fiction foresaw a society where there are no crimes because all the potential criminals have been locked up. In 1949 the writer George Orwell had foreseen something similar in his book 1984, which imagined a world where technology could detect a person’s intention to commit an offence. He called it "thought crime."
Similar thinking has been going on for many years. In the 1960s psychiatrists in the United States recommended that all youngsters should be tested for delinquency with those failing the test locked up in detention villages. Indeed, universities in many countries have done research into anti-social and criminal behavior and tried to create systems to predict crime. The University of California, Los Angeles even had a "Center for the Long-Term Study of Life-Threatening Behavior" with the aim of predicting crime.
Again in the United Kingdom the government is working on a "homicide prediction project" with the fantastical hope that murders can be prevented. In America new AI software is being employed to identify the risk that criminals will re-offend – recidivism – by analyzing not just previous records but also a prisoner’s personal features, ethnicity, schooling, and character.
And of course the prospect of using AI to identify terrorists before they commit mass murder is especially appealing to specialist counter-terrorism law enforcement agencies.
Where will this lead is anyone’s guess at this early stage when AI is in its infancy, but there is a real possibility traditional criminal justice as we know it will be turned on its head. It is a dystopian future where suspected criminals are punished before they commit crimes and suspects are arrested before they break any law.
If that is so, crimes may become non-existent because pre-crime technology has eradicated them so successfully that no one dares to commit a crime. There would be no complaints.
Cheng Huan is an author and a senior counsel who practices in Hong Kong
