San Jose State community members and tech experts say most police departments, including UPD, use artificial intelligence technology for many aspects of the job.
UPD operates a public safety video surveillance system to complement its anti-crime strategy, which helps allocate and deploy personnel and security in public areas, according to its 2020 Policy Manual.
Among the technologies used at SJSU are automated license plate readers, gunshot detection systems and facial recognition, according to the same UPD manual.
In 2017, SJSU officials announced the addition of at least 10 gunshot detectors that can identify loud noises including gunshots and pinpoint the location of the sounds, according to an Oct. 24, 2017 Mercury News article.
Matthew Guariglia, who works with the Electronic Frontier Foundation, said gunshot detection involves microphones activated by gunshots.
Electronic Frontier Foundation is an organization that is dedicated to user privacy and digital freedom, according to its website.
Guariglia said the detectors approximate the location of the gunshot and send the sound to audio experts, who then verify if it’s a gun, which leaves room for human error.
“From what we can tell, they don't record all the time but if a loud noise sets them off, if you are standing in a parking lot and a car backfiring goes off because they're very loud cars . . . what you say near that microphone in the seconds or even minutes after that car backfires is probably going to be recorded.” he said in a Zoom call.
Ethan Dodge, a cyber security and tech expert who’s a San Jose Spotlight journalist, said SJSU uses equipment provided by V5 systems.
V5 systems is a technology company that delivers a security solution that can be deployed in outdoor environments without having to use fixed power or modify existing land or structures to access power and connectivity, according to its website.
“What this gunshot detection technology does is it places microphones within a geographical area and listens for loud noises and then the artificial intelligence tries to determine whether or not it was a gunshot,” Dodge said in a phone call. “If it does determine it was a gunshot then it alerts police and they can respond to that but the problem is it’s often incorrect.”
He said in any situation, there may or may not be a gunshot but the police come to the scene and don't know if someone fired a gun.
Dodge emphasized racial bias within the algorithms of artificial intelligence is another issue within police technology.
He said software and technology “almost always” has the biases of its creators.
“At the end of the day when we look at the criminal legal system, we know that people of color have been prosecuted for crimes at a disproportionate rate as opposed to white people,” Dodge said. “So if the data is biased going in, then the results are going to be biased going out.”
At SJSU, Asian Americans are 34.7% and Hispanic and Latinx are 27.2% of the student body, making up the largest demographic groups, according to the university’s Fall 2020 student quick facts.
SJSU lecturer for “Storytelling in the Age of Intelligent Machines” Nick Stojanovic said there should be more transparency around how some of the algorithms are created and used.
“I would like to see more agency on our part as civilians . . . perhaps more regulation around how that data is used and transparency about how that data is used in terms of storage, sharing and other ways they could potentially use it as well,” Stojanovic said in a phone call.
He said the key feature of policing is judgment.
“One of the main criticisms of where artificial intelligence is right now is that they don’t possess the level of judgment that we do as humans,” Stojanovic said. “There’s a level of consensus that the level of judgment isn’t there and we can’t really ascribe that to the sort of autonomous policing entities that might be out there.”
SJSU engineering professor Ahmed Banafa echoed that sentiment.
Banafa said there need to be laws and regulations to limit the scope of such companies and who has access to the data.
“There is always this challenge of security and privacy, more security, less privacy and vice versa,” he said in an email. “Companies and governments need to find the sweet spot where both meet without overshadowing each other.”
Matthew Guariglia said most people are aware of surveillance technology because they’re surrounded by it and he would like them to find hope in those who are actively fighting back against police using the technology.
“You know, we joke about ‘hey, mister FBI person listening to this call.’ You know, we make these jokes and we call it privacy nihilism [or] the idea that like, our privacy is already so totally gone so why care about it at this point?” Guariglia said. “But the point is . . . if privacy is already gone, why are [governments and companies] trying so hard to take what's left of it away?”
