Artificial Intelligence (AI) is everywhere and if people are not working on it, they are at least reading, using or talking about it. So, before we go into this topic, it is important to define AI. Our friends at Google show the following definition for AI:
“the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages”
We utilize AI at Knightscope. It plays a big part in how we help keep people safe and how our relentless team hopes to one day make the United States of America the safest country in the world (you can read about securing our country here).
We use machine vision, speech recognition and related capabilities. Since we have been in business for over 6 years, we have heard it all: “robots are following me”, “that robot can read my mind” and many more so for this blog we are going to shine some light on what AI can and cannot do so keep on reading!
Myth # 1: “Robots can read my mind”.
Fact: Robots can see, feel, hear and smell aided by sensors on the robot. Sensor information is then sent to the cloud to be processed. There are no sensors available (yet) to read your mind and the ones being worked on (not by us) kinda need to be….ummm….attached to you.
But no, the mind is not something the robot can read today and even if it could read minds, that would just be too much information to send to our cloud servers. Imagine if we are on 4G LTE and there are tons of people around – so much cellular data usage!
Myth # 2: “The robots are dumb – just a camera on wheels.”
Fact: Ahhh….underestimating the technology: not a good idea! Although the Knightscope security robots may not be able to read your mind, they are capable of doing a lot more than video streaming and several folks have found that out the hard way. Our robots are responsible for putting people in jail, including one unfortunate gentleman with 3 felony counts whose bail was denied while awaiting sentencing.
So some bad guys fail to imagine the breadth of capabilities artificial intelligence influences – and they end up in handcuffs. While other potential nefarious actors believe what they see in movies and on TV and think robots are already reading their minds, fluent in over six million forms of communication, or whatever else they might imagine. That in itself helps deter negative behavior. We do drop new software every 2 weeks and new hardware a couple of times a year. The machines will keep getting smarter and smarter and more capable over time. It’s probably not in one’s best interest to chance which feature may ultimately lead to one’s incarceration.
Myth # 3: “The robot is following me!”
Fact: Robots are setup to traverse a pre-defined geo-fenced location (that we call a “deployment”). Robots take different paths to get to different places (especially if there are vehicles or people stop to take selfies with the robot) so paths might not always be the same but the area is predefined, it can be as small as the inside of a mall or as large as a several miles for our larger clients. Think of it as a well defined ‘sandbox’ the robots are confined to operate.
If one of our specialists takes control of the robot remotely, they could get it to follow someone, but this would happen only in the case of a real emergency. Or perhaps in the future, we could have an intentional ‘follow me’ feature for escorting someone to their car, for example. Well, OK, I guess technically doable but is not happening today.
Myth # 4: “This is the technology from Minority Report!”
Fact: Our technology is pretty advanced, yes, and we have helped our customers and the police apprehend suspects and solve crimes but unlike the movie, there are no precogs (living beings who can see future crimes and alert humans) here.
No one sees the future here at Knightscope (well, no one we are willing to identify).
Myth # 5: “The AI will take over and ruin my life. Did you not watch Terminator?”
Fact: At Knightscope, we use what is called Narrow AI, which is AI able to handle just one particular task very well. Think of your email spam filter. We are very thankful for no longer getting so much junk or spam emails anymore but we surely cannot ask the same algorithm to drive an autonomous car as it knows nothing about that other task.
We use computer vision which is based on one of the subsets of AI called Machine Learning. We feed an algorithm images of what it needs to look for and also what it doesn’t, and this algorithm then learns and becomes pretty good at recognizing things. Pretty neat, isn’t it? But there is still plenty of work in the field of AI to even come close to understanding or doing what a human understands or does. Long, long ways from now…
For the foreseeable future, the (friendly) robots are here, and here to stay. But they are here to help humans and in some cases make them superhumans – like we hope to do with our fully autonomous security robots – providing the 2+ million law enforcement and security professionals new ‘really smart eyes & ears’ for them to do their jobs that much more effectively. And actually have a chance to secure 328 million people across 50 states. Well, I guess that makes the point – the real myth in our society is that 2 million brave women and men in uniform are able to secure the country by themselves with no real advanced tools. We aim to fix that!
P.S. One last thing. When citing science fiction movies to us….always remember the emphasis is on fiction!