The secret to AI gunshot-detection accuracy is finally unfolding


This week, New York City’s comptroller published a similar audit of the city’s ShotSpotter system, showing that only 13 percent of alerts generated by the system over an eight-month period could be confirmed as shootings. Auditors noted that while the NYPD has the information it needs to publish data about ShotSpotter’s accuracy, it does not do so. They called the department’s accountability measures “inadequate” and “not sufficient to demonstrate the tool’s effectiveness.”

Champaign and Chicago have canceled their contracts with Flock Safety and Soundthinking, respectively.

“Raven is more than 90 percent accurate at detecting gunshots, while being about equally accurate at detecting fireworks,” Josh Thomas, Flock Safety’s senior vice president of policy and communications, told WIRED in a statement. “And importantly, Raven alerts authorities to incidents of gun violence they would never have known about. For example, in the San Jose report, of the 111 true positive gunshot alerts, SJPD said only 6 percent resulted in a call to 911.”

Eric Piza, a criminology professor at Northeastern University, has conducted some of the most detailed studies available on gunshot detection systems. In a recent study of shootings in Chicago and Kansas City, Missouri, his team’s analysis showed that police responded faster to gunshot incidents, stopped their vehicles closer to the scene of the shootings, and collected more ballistic evidence when responding to automated gunshot alerts than to 911 calls. However, there was no reduction in gun-related crime, and police were no more likely to solve gun crimes in areas with gunshot sensors than in areas where those sensors were not. That study examined only confirmed shootings; it did not include false-positive incidents where the system incorrectly identified gunfire.

In another study in Kansas City, Piza found that reports of shootings in areas with gunshot sensors were 15 percent more likely to be classified as unfounded than reports of shootings in areas without the systems, where police would have relied on calls to 911 and other reporting methods.

“If you look at the different targets of the system, research shows that (gunshot detection technology) generally speeds up police response times,” says Piza. “But research has consistently shown that the number of gun violence victims does not decrease after gunshot detection technology is introduced.”

The New York City Comptroller recommended the NYPD not renew its current $22 million contract with Soundthinking without first conducting a more thorough performance evaluation. In response to the audit, the NYPD wrote that “not renewing ShotSpotter services could put the public at risk.”

In its report, San Jose’s Digital Privacy Office recommended that if the police department wanted to continue using the Raven system, it should continue looking for ways to improve accuracy.

Pointing to the report’s conclusion that only 6 percent of the confirmed gunshots detected by the system were reported to police via 911 calls or other means, police spokesman Sergeant George Garibay tells WIRED that SJPD will continue to use this technology. “This system is still proving useful in providing supplemental evidence for a variety of violent gun crimes,” he says. “Hopefully more crimes will be solved and arrest efforts will increase which will lead to a reduction in gun violence.”


Please enter your comment!
Please enter your name here

Share post:




More like this

Here are all the devices compatible with iOS 18

Apple's WWDC 2024 was full of announcements about iOS...

Will Smith broke Twitch’s biggest streaming record, the real reason why

Every summer, Spanish Twitch streamer Ibai Llanos hosts a...