Predicted headline date: 21/06/2019
In an astounding development, the San Francisco Police Department arrested a murder suspect based on a photo identikit created from a DNA sample found at the crime scene. Once the photo of the subject was created from the DNA sample, it was distributed to inhabitants of the Richmond District via Social Media. An anonymous caller identified the suspect close to the corner of 22nd Ave and Clement Street. The police was notified last Friday and the suspect was apprehended. A comparison between the photo identikit and the suspect’s picture had a 93% match rating of dominant facial features.
Police Superintendent Lykes Donutt was predictably upbeat in his statement to local and international news media. “The photo identikit was developed by our new Computational Social Science department”, he boasted. “The latest Machine Learning techniques were used to make an accurate prediction of the suspect’s facial features, hair, eye-colour and age.” Superintendent Donutt used the media exposure to release five new photo identikits of suspects in other crimes, also based on DNA samples collected at various crime scenes. He believes that if this trial is successful, a number of other Police Departments will follow suit.
Anonymous sources in the Police Department said that this initiative has been months in the making. Ever since personal DNA analysis became a must-have for individuals to manage their health and diet, companies like 23andMe and Ancestry .com processed millions of DNA samples in preceding years. The sites also offered to use your DNA to link your profile with unknown relatives, and the opportunity to link a photo to your DNA profile. In the process, hundreds of thousands of members gave permission for the anonymised data to be shared. It was this initiative that opened the door for what transpired today. A variety of Machine Learning tools were used to work through the multitude of pictures and DNA data points to try and determine patterns between the DNA and associated facial features. Within a short period of time, the DNA strains that link to specific facial features have been identified. An algorithm for facial feature prediction was launched by the University of Stanford in October 2018, which lead to the interest of the SFPD shortly thereafter.
The lawyer for the suspect, Hylie Suspisus, is adamant that there were no grounds for his client to be arrested. Photo identikits have traditionally been created by eyewitnesses to the crime. In this case, there was no eyewitness account, and the photo identikit could therefore not be used to arrest his client. He is adamant that the case will be dismissed in his favour. Asked if the suspect is prepared to be interrogated by JAMES3000, the Machine Learning bot that looks at micro-expressions and brain-scans to determine if a suspect is lying, Mr Suspisus declined. Although JAMES3000 has a 99,999% record in correctly measuring brain activity, Mr Suspisus said that guilt should not be established via the search for subjective truth but rather through the correct application of legal precedent.