Who Says Killer Robots Are Bad?

I, for one, welcome our new overlords. Just wanted to get that out there. You know, just in case. So, Patty J says I should talk about robots in our everyday lives a bit for you, and here goes.

I hate cleaning up. I really do. Cleaning cat barf, making the bed, taking out the trash. Wouldn’t it be cool if I had a robot companion that would do all those things for me so I could watch Netflix in peace?

Let's start with robots in general. Asimov (I, Robot, 40’s and 50’s) and the Sci-Fi people were writing about this stuff back in the 1930’s and it was a very popular topic indeed. From Robbie the Robot (Forbidden Planet, 1956), and Rosie, the Robot Maid (Jetsons, 1962-3), to HAL (2001: A Space Odyssey, 1968 movie based on A. Clarke's book 1968), Roy Batty (Blade Runner, 1982), and Robot Companions (sex robots, present day), the list goes on and on. We are fascinated with the idea that we could have these artificial things around that do all sorts of tasks for us.

Asimov talked a lot about the "laws of robotics" which prevented robots from harming humans, but Terminator (1984) really presented us with a quandary. What if robots were "intelligent" and, as such, decided that we were more of a pest then a benevolent master? When you think of the poor sex robots (whether modern or poor old Priss in Blade Runner) well, that kind of makes sense. Today, we are starting to see some of these ideas in practice.

Now, the idea of a robot was originally, just basically an "automaton" which could do a task over and over that it learned from some instruction. Say, moving a mop back and forth. Ok, well that's useful in simple tasks, but humans can do what are called "fuzzy tasks": A fuzzy task is something where the objective is NOT so clear as putting a bolt on a thread. Suddenly, the idea of Rosie the Robot maid became clearer. What if you could build something that could take a fuzzy instruction like "clean" and adapt that based on past skills just like humans do. Humans learn by observation and trial and error - You touch something hot as a child and you don't want to touch it again since it damaged you. An AI robot could do that and more. So, Rosie could learn to "clean" by watching Jane Jetson clean up and then take over. (I know it's sexist, but it was 1962, what can you do?)

These types of robots are already all over the place. Stop and Shop now has a robot patrolling the aisles "for your protection.” Those take pictures of people in the store and their purpose is primarily to look for spills, hazards, who knows what all and they can perform "fuzzy" tasks. Probably they will soon be cleaning up on aisle 7 when someone drops a jar of mayonnaise. Likewise, we are starting to see drones that can "follow while flying" all on their own. I did it last year. The drone flew behind the car, despite high winds and other challenges without any input from the operator. It was just constantly adapting and solving that complex problem. Even that little Roomba robot in your house learns the layout, figures out there is cat barf on the floor and mops it up. Wow. 

So, what about the sex robot. (Yuck.) Well, they have them. Numerous companies are rolling out male and female models that can talk, react, you name it. Why not? People have sought companions that appeal to them since two bacteria said hello in the primordial soup. Why not a device that can talk to you, try to make you happy, whatever. Probably more than just a sex thing. Now, all this is AI. That means basically something that emulates a human, learns, and performs tasks for you. Let's add another idea, Machine Learning.

ML is something else entirely. Machine Learning means something that can exceed its original specs. Let's stick with sex for a minute. What if the sex robot can not only be your companion, but also learn and grow in some other way? Maybe the ML sex robot doesn't just say "Hey, how you doin'" and decides to go to the internet and learn a new trade. ML means this "thing" may learn in ways we don't understand or maybe can't even hope to understand. It's kind of like a super genius who sees the world in a new way. Cool, but also scary to us poor humans. 

A lot of this discussion centers around ethics and morality. Now, humans have been working on these concepts for thousands of years to no avail. Some people think that only through religious doctrine can we be "good".  But there are plenty of atheists running around living normal lives. So, that kind of shoots down that theory. We may not be able to understand the ethics of ML and, thus, we may end up in the Terminator world where Skynet (the ML in that case) decides to get rid of the human pests. Or, we could just end up second fiddle as they say. Where poor inferior humans end up just kind of hanging around wondering what the ML is up to that has gone way beyond our own poor understanding. All pretty scary, but then change always is.

If you want to remember the difference in AI and ML just remember this:  AI is a spell/grammar checker, but ML does your rewrites.

Even with just AI type robots like say Metalhead (Black Mirror, 2017), humans could end up in a jam when the security robot which was supposed to protect us, well, protects us a little too well. Asimov was trying to deal with that back in the 40’s by building in restrictions to the AI. Blade Runner had ML robots that automatically died after a period of time to prevent them from getting too smart, and on and on. Hard to say where this ends up.

I don't know what the future holds with personal robots. I think the AI trend will continue because we really like convenience. I want an autonomous car so I can sit back and read a book while it drives me to work.  That's good AI. I think those AI devices will outdrive most of the people I know and see on the road. I would love to have a smarter Roomba that will clean my house thoroughly while I watch Netflix. Sounds good. I want one smart enough to leave my cats alone, but still clean up and make the beds even if I didn't tell it to do so ("Hey, Rosie, get over here."). I would love to have had a micro drone swarm that surrounded my daughter when she went to school. That swarm would make sure that if a rabid dog came around, it would get chased away, or maybe when some sex predator stepped out, the swarm could say "back off" and take care of her.  All those things are almost right here, really. AI is taking leaps and bounds and you are going to see some amazing things in the next 20 years. A smart house that adjusts temperatures, reports damage, maybe even fixes the damage on the fly, all right here and now.

So, one more time with the sex robot (I know, I couldn’t resist). The base idea is kind of disgusting but what if it was a companion like Sherman (Millennium, 1977, 1989, by John Varley) who could be your pal? I kind of like that idea. Your perfect friend who is a confidant, priest, sex robot, you name it. With AI, well, it's kind of limited but doable. With ML, Sherman could become all those things and more. I just hope my robot buddy doesn't decide I am kind of a jerk and hit the road leaving me back where I started cleaning up cat barf. Hey Rosie, get over here.

More About Doug White

Doug White is the Chair of Cybersecurity and Networking programs at Roger Williams University. He has worked in the technology industry for many years and specializes in networking, disaster, forensics, and security. He has been paid to break into buildings, talk tech people out of their usernames and passwords, steal money, and figure out horrible scenarios like “What is if a rabid shark swarm was caught up in a tornado while a core meltdown occurred? Could we still watch Netflix?” Doug has a PhD in Computer Information Systems and Quantitative Analysis from the University of Arkansas, is a Certified Computer Examiner, A Cisco Certified Network Administrator, A Certified Information Systems Security Professional, and a licensed private investigator.