AI Can Help Solve Unsolved Crimes, Here's How
Wed, April 21, 2021

AI Can Help Solve Unsolved Crimes, Here's How

If you end up murdered in the US, there’s a 1 in 3 chance that the police won’t identify the person who killed you. Unsolved crimes like this are nothing new / Photo by: Openfinal via Shutterstock

 

If you end up murdered in the US, there’s a 1 in 3 chance that the police won’t identify the person who killed you. Unsolved crimes like this are nothing new. Most of us are familiar with cold cases, from murders, disappearances to theft. 

Since the 1960s, there have been over 200,000 unsolved murders in the US. A 2018 FBI report reveals that roughly 40% of the country’s murders went unsolved in 2017. The authorities looked at 2017 crime and arrest data from 16,000 law enforcement agencies across the US and found that only 61.6% of reported murders were “cleared.” Since 2009, law enforcement’s success rate has not improved. 

A 2017 report by Pew Research, a nonpartisan fact tank that informs the public about the issues, attitudes, and trends shaping America and the world, most of the reported violent and property crimes that occur in the US don’t result in the arrest, charging, and prosecution of a suspect. 

According to Business Insider, a fast-growing business site with deep financial, media, tech, and other industry verticals, one of the reasons this happens is because homicide investigations are expensive. There are some states, like Detroit, that can’t afford to decrease investigators' caseloads. Vernon Geberth, a retired and self-described NYPD "murder cop,” stated that the public doesn't realize that clearing murders has become harder in recent decades. The standards for charging someone for a particular crime are too high now.

Geberth thinks that prosecutors are demanding that police deliver "open-and-shut cases" that will lead to quick plea bargains. While modern tools like DNA recognition have helped, the relationship between police and the public has worsened due to these unsolved crimes. "If there is a distrust of the police themselves and the system, all of these scientific advances are not going to help us," he said. 

Solving Crimes Through AI

The rising cold cases don’t bring justice to the family and loved ones of the victims. But beyond that, unsolved crimes means leaving an increasing number of killers out on the streets, undermining safety in urban neighborhoods, and eroding confidence in the criminal justice system. Fortunately, artificial intelligence can help. 

Earlier this year, researchers from Northumbria University released a study emphasizing how AI can crack unsolved crimes. It can also provide insight into the weapons used in committing the crime. The team believes that this approach can solve the case of the Bloody Sunday killings of 1972, where 13 people were killed and 15 people were wounded. Dr. Matteo Gallidabino, Senior Lecturer in Forensic Science at Northumbria University, stated that the investigators found large amounts of gunshot residue (GSR) all over the victims.

While it has been concluded that the killings resulted from shootings, it was later established that they were likely caused by secondary, post-event transfer of contamination from military staff. “Small amounts of GSR, indeed, may be transferred by prolonged contacts with contaminated surfaces, such as those that took place when soldiers helped transport victims to the hospital after the event,” Dr. Gallidabino said.

According to Phys.org, an internet news portal that provides the latest news on science, the team used machine learning for this study. Dr. Gallidabino stated that they had modeled complex data relationships with the help of machine learning algorithms. "Through careful fine-tuning, these can be applied to predict important characteristics of the ammunition used in a particular shooting event from those of the respective gunshot residue (GSR) deposited on surrounding surfaces or items, such as spent cases, wounds and, potentially, also the shooter's hands,” he explained.

The researchers believe that their findings are an improvement from current methods of GSR analysis. Their approach has unprecedented accuracy, which is said to bring fresh insights to some high profile, unsolved crimes of the past. Dr. Leon Barron, Senior Lecturer in Forensic Science from King's College London, stated that the combination of lab analysis and computer-based machine learning will enable experts to capitalize on the vast amounts of data. These data can be generated to make ground-breaking advances like this approach. 

Earlier this year, researchers from Northumbria University released a study emphasizing how AI can crack unsolved crimes. It can also provide insight into the weapons used in committing the crime / Photo by: Robert Kneschke via Shutterstock

 

Developing and Testing the Approach

According to Medium, an online publishing platform, the team used machine learning for forensic applications in this study. They collected data from firing a range of ammunition including gun cartridges and smokeless powders. This was meant to help the team figure out the relationship between the ammunition and the residue. 

Dr. Gallidabino analyzed the gun cartridges as well as the volatile part of the GSR before focusing on the original smokeless powders. He then established the relationship between the ammunition and the residue. The good news about this new approach is that it can be widely used in the field of forensic science and analytical chemistry. 

"The benefits are countless. They may even extend to other fields in analytical sciences that routinely encounter changeable chemical traces, such as the analysis of improvised explosive devices, arson accelerants, and environmental pollutants,” Dr. Gallidabino said. 

With AI and machine learning, the cases that have long been forgotten have a possibility of being solved. This opens a lot of opportunities for future researchers to design new approaches in the field of crime.