Do Security Robots Reduce Crime? (nbcnews.com) 50
Westland Real Estate Group patrols its 1,000-unit apartment complex in Las Vegas with "a conical, bulky, artificial intelligence-powered robot" standing just over 5 feet tall, according to NBC News. Manufactured by Knightscope, the robot is equipped with four internal cameras capturing a constant 360-degree view, and can also scan and record license plates (as well as the MAC addresses of cellphones).
But is it doing any good?
As more government agencies and private sector companies resort to robots to help fight crime, the verdict is out about how effective they are in actually reducing it. Knightscope, which experts say is the dominant player in this market, has cited little public evidence that its robots have reduced crime as the company deploys them everywhere from a Georgia shopping mall to an Arizona development to a Nevada casino. Knightscope's clients also don't know how much these security robots help. "Are we seeing dramatic changes since we deployed the robot in January?" Dena Lerner, the Westland spokesperson said. "No. But I do believe it is a great tool to keep a community as large as this, to keep it safer, to keep it controlled."
For its part, Knightscope maintains on its website that the robots "predict and prevent crime," without much evidence that they do so. Experts say this is a bold claim. "It would be difficult to introduce a single thing and it causes crime to go down," said Ryan Calo, a law professor at the University of Washington, comparing the Knightscope robots to a "roving scarecrow." Additionally, the company does not provide specific, detailed examples of crimes that have been thwarted due to the robots.
The robots are expensive — they're rented out at about $70,000-$80,000 a year — but growth has stalled for the two years since 2018, and over four years Knightscope's total clients actually dropped from 30 to just 23. (Expenses have now risen — partly because the company is now doubling its marketing budget.)
There's also a thermal scanning feature, but Andrew Ferguson, a law professor at American University, still called these robots an "expensive version of security theater." And NBC News adds that KnightScope's been involved "in both tragic and comical episodes." In 2016, a K5 roaming around Stanford Shopping Center in Palo Alto, California, hit a 16-month-old toddler, bruising his leg and running over his foot. The company apologized, calling it a "freakish accident," and invited the family to visit the company's nearby headquarters in Mountain View, which the family declined. The following year, another K5 robot slipped on steps adjacent to a fountain at the Washington Harbour development in Washington, D.C., falling into the water. In October 2019, a Huntington Park woman, Cogo Guebara, told NBC News that she tried reporting a fistfight by pressing an emergency alert button on the HP RoboCop itself, but to no avail. She learned later the emergency button was not yet connected to the police department itself... [The northern California city] Hayward dispatched its robot in a city parking garage in 2018. The following year, a man attacked and knocked over the robot. Despite having clear video and photographic evidence of the alleged crime, no one was arrested, according to Adam Kostrzak, the city's chief information officer.
The city didn't renew its contract "due to the financial impact of Covid-19 in early 2020," the city's CIO tells NBC News. But the city had already spent over $137,000 on the robot over two years.
For its part, Knightscope maintains on its website that the robots "predict and prevent crime," without much evidence that they do so. Experts say this is a bold claim. "It would be difficult to introduce a single thing and it causes crime to go down," said Ryan Calo, a law professor at the University of Washington, comparing the Knightscope robots to a "roving scarecrow." Additionally, the company does not provide specific, detailed examples of crimes that have been thwarted due to the robots.
The robots are expensive — they're rented out at about $70,000-$80,000 a year — but growth has stalled for the two years since 2018, and over four years Knightscope's total clients actually dropped from 30 to just 23. (Expenses have now risen — partly because the company is now doubling its marketing budget.)
There's also a thermal scanning feature, but Andrew Ferguson, a law professor at American University, still called these robots an "expensive version of security theater." And NBC News adds that KnightScope's been involved "in both tragic and comical episodes." In 2016, a K5 roaming around Stanford Shopping Center in Palo Alto, California, hit a 16-month-old toddler, bruising his leg and running over his foot. The company apologized, calling it a "freakish accident," and invited the family to visit the company's nearby headquarters in Mountain View, which the family declined. The following year, another K5 robot slipped on steps adjacent to a fountain at the Washington Harbour development in Washington, D.C., falling into the water. In October 2019, a Huntington Park woman, Cogo Guebara, told NBC News that she tried reporting a fistfight by pressing an emergency alert button on the HP RoboCop itself, but to no avail. She learned later the emergency button was not yet connected to the police department itself... [The northern California city] Hayward dispatched its robot in a city parking garage in 2018. The following year, a man attacked and knocked over the robot. Despite having clear video and photographic evidence of the alleged crime, no one was arrested, according to Adam Kostrzak, the city's chief information officer.
The city didn't renew its contract "due to the financial impact of Covid-19 in early 2020," the city's CIO tells NBC News. But the city had already spent over $137,000 on the robot over two years.
Tragic? (Score:2)
The only thing tragic in the summary was it ran over a kids foot.
Impossible to know is how much crime was deterred by the pretense of these robots, even if all they are doing is recording that would deter some thieves.
I would say they are actually doing pretty well if the incidents listed are the only problems encountered with the robots.
They seem like a really good idea, but they really need to get that price down as for that rate a place could have a private security guard...
Re:Tragic? (Score:5, Interesting)
Impossible to know is how much crime was deterred by the pretense of these robots
Why is that impossible to know?
An obvious way to find out is to put them in some malls but not others and then compare the results.
Another obvious experiment would be to have them patrol on some days but not on others. If they work, there should be fewer reported incidents when they are patrolling.
Re:Tragic? (Score:4, Informative)
The potential criminals would not be aware on which days the robot is going to be present, so knowledge that they *might* be present could also deter crime, or shift the criminals to other malls.
The number of reported incidents could increase by virtue of incidents being easier to detect/report rather than an actual increase in incidents. Many crimes go unreported/undetected.
Re: (Score:3)
Which is why companies like this exist.. they are not really selling crime prevention, they are selling anxiety reduction devices.
Re: (Score:2)
Why is that impossible to know?
An obvious way to find out is to put them in some malls but not others and then compare the results.
Another obvious experiment would be to have them patrol on some days but not on others. If they work, there should be fewer reported incidents when they are patrolling.
The control case needs to be selected carefully. Do we replace the robot with human security? Do we have no security at all? Frankly, I would like to see all three cases tested. What impact does a security officer have on crime in the first place? Is a security robot comparable in effectiveness? The results might be surprising.
Yeah I'd choose the human for sure (Score:3)
> They seem like a really good idea, but they really need to get that price down as for that rate a place could have a private security guard...
Indeed. A base-level security guard isn't a high-paying job (high cost). Figure the cost is 50% more than the employee wages. The person is cheaper than the robot, and smarter.
Re: Yeah I'd choose the human for sure (Score:3)
WelÃ, ED-209 wasn't smart, but pretty effective.
You can't do security by buying a *thing*. (Score:4, Interesting)
As much as product marketers try to sell that way, security is a *system* property. But that doesn't mean *things* can't be useful in a well-thought-out system.
My problem is that $70,000 would buy an awful lot of security cameras, and unlike a security robot with just one camera they can be everywhere at once. You need people watching the cameras of course, but that's where the AI can come in; flag stuff that a person needs to review.
Re: (Score:2)
$70,000 would buy an awful lot of security cameras
$70k wouldn't pay the burdened salary of even a single security guard.
The security guard only works 8 hours/day, has weekends off, and takes vacations.
Re: (Score:3)
That would be a compelling argument if you thought the robot is capable of replacing a security guard.
Re: (Score:1)
I have seen those security robots slowly get better. The thing the robot gives is a presence. Yes, people can sack the robots, but that means someone goes from trespass to a high felony for malicious mischief. If one detects someone or a banned license plate, the fact that it pops off a siren means someone has to do something, and that something is likely getting the heck out of Dodge.
For buildings and level grounds, they are a great force multiplier, and if one detects someone armed, it can allow securi
Re: You can't do security by buying a *thing*. (Score:2)
How will the $ 80000 robot handle US-specific crimes like BurnLootMurder (BLM) riots?
Re: You can't do security by buying a *thing*. (Score:2)
70k sounds a lot like privilege to me.
Obviously they arenâ(TM)t putting Robertcop into the projects.
Re: (Score:2)
I wonder how effective the robots are. There was a product a while back that was basically a laser light show, designed to make criminals think you had some kind of high tech security system like out of a movie.
Re: (Score:2)
My guess would be half a year until it is common knowledge in those concerned.
Re: (Score:2)
You can't use AI to monitor crime because as soon as you notice certain patterns arise in certain demographics you get called racist and brought before a kangaroo court. Because what to a normal thinking person might be evidence of gang affiliation, i.e. an ethnic group favoring certain symbols and primary colors for example, must be ignored to the point of enacting violence on the observer who calls attention to it by the retarded tumbler generation. And whilst a security guard can keep this knowledge to h
Re: (Score:2)
Your position illustrates the problem with AI systems: people are too credulous. They jump the conclusion when the system identifies a pattern that that pattern is real when it can just as well be an artifact of the training data set.
If a system categorizes the *same actions* by a black subject different than a white subject, that's a flaw. There aren't -- or at least shouldn't be -- different laws for white people and black people.
Re: (Score:2)
I'm speaking more on the predictive side. If you see a large group of 18-30 year old's walking in close proximity to each other, all wearing blue for example and on the other side of the mall there is a similar group all wearing red; you would have to be an absolute moron not to preemptively call the police. At the same time, there is so much in fighting among these groups that an AI isn't going to realise that one group wearing red and black is effectively at war with the another group wearing black and re
Re: (Score:2)
There seems to be less focus on Why the person is committing the crime, and focus on finding and hunting down the person who had committed it.
There is popular opinion that most of the crimes are from psychopaths, who really have no morals, or empathy and will do whatever they can get away with it just because they want to.
While the psychopath are out there, and they are a problem. A lot of the crime out there is often from a more complex set of issues. Often the criminal feels they are in a situation with
Re: (Score:2)
The UK has pretty much demonstrated that cameras are useless to reduce crime.
Futurama (Score:3)
If 140 episodes of Futurama have taught me anything, they've taught me the answer to this question is a resounding "Ooohh yeahh", uh, no.
All Hail Hypnotoad!
Seems like a valuable piece of equipment. (Score:4, Interesting)
Re: (Score:2)
Since 2020, iOS 14 / WatchOS 7, Android 10 are all supposed to feature MAC randomization to prevent exactly this kind of tracking.
No more so than security cameras (Score:3)
London has cameras on every street corner (Score:2)
And yet, even with all the crime being reported, that adds a certain degree of comfort that it will be SOLVED....
But if experience is any teacher, it doesn't prevent crimes of passion, crimes of opportunity, or pre-planned crimes. So basically, it only stops people who are afraid to get caught. Those people generally dont commit crimes anyway, so I believe the effect is very limited.
That said, even a limited or even a perceived effect, can be enough that some people want it. London being a prime example. At
Security Robot...from Dr. Who...Blake's7 (Score:2)
A security robot from Dr. Who, like the ones from "The War Machines" or better yet, a Dalek. :) When you're trespassing and a robot says "Exterminate!" you run away.
I liked the Federation security robot from "Blakes'7" where if it found intruders, a hatch opened, and a flamethrower sprayed flame over the miscreants.
Of course the movie "Runaway" with all the crime going on, the security robots did a poor job.
A security robot is taking a security camera and making it mobile...and adding new problems that come
Re: Security Robot...from Dr. Who...Blake's7 (Score:1)
I just go down the stairs . Or tip it over
They need good background investigations (Score:2)
Re: (Score:2)
What the rich & influential do is often overlooked ... the rest of us are examined for criminal activity (as well as pay our taxes).
EXTERMINATE! EXTERMINATE! (Score:4, Funny)
Instead of increasing its marketing budget, Knightscope should license sound effects and likenesses from the BBC. A constant presence of Dr. Who fans with mobile phones taking pictures will prevent more crime than the robots by themselves.
No (Score:2)
Re: (Score:2)
Lol how is this robot not already stolen and scrap (Score:1)
Itâ(TM)s got to be worth an easy $200
Um (Score:2)
[The northern California city] Hayward dispatched its robot in a city parking garage in 2018. The following year, a man attacked and knocked over the robot. Despite having clear video and photographic evidence of the alleged crime, no one was arrested, according to Adam Kostrzak, the city's chief information officer.
This is rattled off as though it were a failure of the robot. In reality, the robot gathered the evidence, but the humans failed to act on it, is what it sounds like.
Sort of (Score:2)
I imagine that they reduce the rate at which criminals get away unpunished.
I think that in theory, it might eventually reduce crime, but it would have to be a well known and established part of society for at least a generation.
Of course they do! (Score:2)
Who would argue with that? If you still don't think robots reduce crime, then you're an anti-American, pro-crime communist! - Pat Novak
Wildly overpriced (Score:3)
You can BUILD very captable robots for way less than 70k. I think Knightscope is just a venture capital exploitation vehicle. At 70k a year of course the economic model doesn't work. Its a camera on a platform. If you set them up as telepresence platforms to leverage highly capable director type guards with a small emergency response group to back them up you can pay people decently, provide solid security and reduce the mindnumbingness of security work.
Re: (Score:2)
70k a year is really not a lot to pay for a failed experiment, though. They could be out a lot more money if they had tried to build or purchase something and it still turned out to be a total bust. Outsourcing it really is the right move, unless you have local expertise on tap.
Too tempting (Score:2)
If I saw one, I would be way too tempted to attach a digital player with a flashing light to the thing so it would randomly scream "EXTERMINATE!".
Immune to mind-affecting (Score:2)
You cannot reason with, bribe, coerce, threaten, or intimidate a robot. You could easily threaten or bribe someone into compliance, but a robot? No chance in hell.
Re: (Score:2)
Betteridges law of headlines says... (Score:2)
Re: (Score:2)
Cheaper & more effective (Score:2)