I’ve recently come across a report titled “ShotSpotter and the Misfires of Gunshot Detection Technology” by the STOP Surveillance Technology Project.
The report is extremely critical of this Acoustic Gunshot Detection Technology (AGDT) suggesting it offers little operational value for the police and it can prove detrimental with serious ethical caveats.
How bad is this technology really?
SHOTSPOTTER®
ShotSpotter Inc. has been on the market for 20 years and it is the leading AGDT on the market used by over 200 police jurisdictions worldwide, with at least 130 of those from the US.
ShotSpotter prides itself with the slogan: “Right place, right time. Precision policing solutions that help save lives, solve cases and deter crime – making communities safer.”
 |
| AGDT technology starts with microphones mounted above the city on buildings and other tall structures |
ShotSpotter combines narrow AI with human experts to identify and locate gunshot events and their location in the neighbourhoods where it’s been installed. Its detection system comprises microphones placed on streetlights, buildings and utility poles, and audio software that compares the sounds with prior gunshot events in its database. The system triangulates the location of the sound that is further reviewed by a human sound analyst who alerts the police should they determine the sound actually derives from a gunshot.
THE GOOD
Company reports, and some technology reviews, offer that it detects shootings that would go unreported, locates gunshots for quick police response, and improves the arrest rates of suspects. This all leads to preventing gun violence and enhanced neighbourhood safety.
THE BAD
The evidence suggests that many of the above claims can be disputed. For example, while quasi-experimental studies indicated over 80% accuracy in detection and triangulation, other studies show the AGDT works poorly in detecting gunshots and, instead, it suffers from what researchers call “false positives” (it gets it wrong).
Chicago’s Office of Inspector General review found that only 9.1% of probable gunshot alerts out of 50,000 records over a 17-month period resulted in evidence of a gun-related offence.
 |
| While we sleep at night, it listens for gunshots from roof-mounted listening devices - what could go wrong? |
These studies conclude that AGDT may actually be ineffective and inaccurate and can thus waste police resources. None of that mentions the cost, (Chicago paid $33 million for 2 years of service).
As well, there are poor outcomes on firearm offence reduction and shooting victims’ medical outcomes two years following the installation. Thus, it did not lead to gun violence deterrence as anticipated.
Most alarming was whether ShotSpotter evidence should be used in court. The company has previously admitted to altering the records and reclassifying the sounds that were initially not detected and determined as gunshots but were later reclassified as such based on police reports. While this may be occasionally warranted, at other times it may lead to the misuse of data and misrepresentation of the evidence on the side of the law enforcement agencies.
THE UGLY
AGDT can lead the local community to become disengaged and stop reporting gunshots because they believe the technology will handle that for them. In St Louis, for example, citizen-initiated reports of gunshots before the installation of ShotSpotter were 7 times more useful than AGDT and wasted less police time. Following the installation, citizen reports significantly decreased and decreased further when the system was expanded.
AGDT is predominantly deployed in poorer and majority Black and Latinx neighbourhoods commonly identified as hotspots for shootings and violence. While there might be more shooting incidents in those neighbourhoods, this automatically means AGDT will detect more shooting incidents in those specific neighbourhoods rather than in other neighbourhoods where it is not deployed. This might perpetuate the problem of over-policing and racial bias in communities of colour.
 |
| Evaluations of AGDT shows problems with "false positives" - it turns out the city can be a noisy place |
There is a similar problem with another narrow AI program called predictive policing where there are reports of racial discrimination and over-policing and the additional concern of “tech washing”.
Despite ShotSpotter’s claim that the sensors cannot record private conversations, there are increasing concerns of the microphones recording regular conversations and acting as a “massive eavesdropping device”.
In one case, such a private conversation in a public space was intercepted and admitted into evidence in a murder trial court case.
Currently, there appear to be no regulations that would prevent ShotSpotter from sharing stored audio recordings.
 |
| SafeGrowth works directly with community residents in collaboration with the police to plan their own safety strategies |
SMARTER APPROACHES
After two decades, the results of AGDT implementation continue to show few positive effects. In fact, some results are detrimental to effective police work, police-community relationships and community safety.
The evidence suggests that police leaders should err on the side of caution and turn towards approaches for addressing gun-related violence that have a better track record.
One example: combining neighbourhood policing and problem-oriented policing to build police-community relationships, act proactively, and approach the problem at its core.
Another that we obviously recommend is SafeGrowth. This blueprint for neighbourhood safety planning helps residents help themselves, in partnership with police and using the latest in crime prevention. SafeGrowth provides a platform to help address gun-related violence and other crime and safety problems.
The documented success story in the New Orlean’s gun-violent neighbourhood of Hollygrove is only one example of how neighbours disrupted gun violence.
Sometimes the most effective solutions are not technology, they are human-based and neighbourhood friendly. That, we believe, is a future we all want.
Post a Comment