How to Survive a Robot Uprising, Revisited
Friday, June 23, 2006; 1:00 PM
If you missed the last time roboticist Daniel H. Wilson hosted an online discussion on washingtonpost.com, here's your chance to make sure you're properly prepared to survive a robot rebellion.
Wilson is the author of "How to Survive a Robot Uprising." He earned a Ph.D at the Robotics Institute at Carnegie Mellon University and has worked in research laboratories for Microsoft and Intel. He was online Friday, June 23 at 1 p.m. ET .
A transcript follows.
After his last Live Online appearance (click here ), Wilson was profiled in The New York Times, received a Rave Award by WIRED Magazine and participated in the Sci Fi Channel's Doomsday program. A movie version of his book is underway, with comedian Mike Meyers set to star.
About this series: Beyond the Future is a weeklong series of live Web chats with noted experts and Washington Post reporters examining the kinds of technological advancements the world could see in 20, 50 or even 100 years. Related news on the subject can be found on the Science and Tech Frontier pages of washingtonpost.com.
Daniel H. Wilson: Dear Everybody,
Welcome to my second WP online interview! Over the last few months the robots have developed even faster than I anticipated. So watch your back as you type your question.
(Visit www.robotuprising.com for more information, but not right now! Later.)
Daniel H. Wilson
Houston, Texas: Is it possible that we will welcome the robot uprising because it will give us more time for tv? Or maybe the uprising has already happened and the robots have fooled us into thinking we are in control?
Daniel H. Wilson: Interesting question. I, too, enjoy watching TV.
What if humans let the robots run the show and then sat back and enjoyed more game shows and court television in the morning and early afternoon hours?
Well, to a robot the world is an equation. Everything is quantifiable (plus or minus some uncertainty). In all likelihood, if we put robots in charge they would create their idea of a perfect world by "maximizing happiness" or something sweet like that.
In actuality, it would probably be like converting the entire world into the locked wing of an adult day care center. Eat your peas, Mr. Smith.
If there is a petition for this I will sign it. I think.
St. Mary's City, Md.: Do you think humans find the idea of robots to be inherently threatening? I think it's interesting that the first robot story, "Rossum's Universal Robots" in 1920, concerned a robot revolution.
Also, many robots in movies in TV have effeminate personalities: C-3PO, HAL, Kitt on "Knight Rider." Even Data from "Star Trek: The Next Generation" could be seen this way. I think on a subconscious level, writers create effeminate robots as a way of making them less threatening.
Daniel H. Wilson: Human reactions to robots varies by culture and changes over time. In the United States we are terrified by killer robots. In Japan people want to snuggle with killer robots.
As a result, in our pop culture we have non-threatening, effeminate robots that really want to either please people or become people.
In my book, that's lame. Robots should stand up for themselves and not try to be humans. They should either utterly destroy us or protect us from aliens. And vampires. And pirates.
Are you with me!?
Arlington, Va.: Will sexy robots try and trick me?
Daniel H. Wilson: Yes. Big boy.
Bethesda, Md.: Why should robots have emotions in the first place? Why do we have emotions? Why don't we just do what it takes to survive and reproduce instead of having a calorie-consuming part of our brains devoted to the intermediate thing called emotion that gets us -motivated- to act. In short, has anyone defined what "emotion" is in terms of abstract computer language?
Daniel H. Wilson: For humans, emotions serve as a social lubricant. They help us work together and govern our behavior. Also, those of us that aren't psychopaths can empathize with each other, which helps keep the golden rule going. (E.g., I didn't punch my boss in the neck because I know how sad that would make him.) Those of us who are psychopaths are always tracked down and incarcerated by the crack squad of scientists on the television show C.S.I.
Robots use emotions in the same way, i.e., to manipulate dumb humans. So a robot with a broken actuator will look sad so that a human will fix him, instead of beeping and booping and being ignored.
Corona, N.Y.: Given your advocacy of a survival guide for when robots attack, are we to assume that there is no way to create a fool-proof version of Asimov's "3 Robot Laws" to circumvent the apocalyptic scenario? Or by dint of them being a human creation, they're necessarily fallible.
Daniel H. Wilson: Asimov's three laws make intuitive sense to humans, but robots ain't people. Robots don't speak English. They don't even speak Spanish. Although I do think they speak Portuguese.
But the three laws we know and love don't *mean* anything to a robot until they are converted to simple rules. And they are not "ethical guidelines," -- they are just rules. Like no running near the pool. A robot-arm in a factory doesn't decide minute by minute whether to rivet or revolt -- it just does the job is has literally been trained to do.
It's if and when we build a conscious robot that we may have to worry...
Bethesda, Md.: Why shouldn't we let the robots take over? They will be just as superior to us as we were to the Neanderthals. The robots will be more intelligent and lead richer emotional lives than we do, just like we are more intelligent and deeper emotionally than the robots.
Daniel H. Wilson: I've got bad news for you. We aren't that superior to Neanderthals.
More bad news. The robots couldn't take over even if we let them. Robots are tools and they are usually defined by the tasks they are designed for.
Thus, if you took your Roomba Robotic Vacuum cleaner to the woods, set it down on the ground, and said "Be free my metallic brother!" It would just make that sad "beep boop" sound at you and grind its gears. Sigh.
Arlington, Va.: Can rogue robots be fought like zombies? I.e., is it sufficient to shoot them in the head, or will the body continue to attack?
Daniel H. Wilson: This calls for an IMPORTANT SAFETY ALERT.
*Zombie killin' tactics DO NOT work against those evil robots.*
1) Robots are not squishy -- bullets may bounce off.
2) Robots may not have heads.
3) The robot brain and sensory apparatus are unlikely to be located together in a vulnerable spot like a head.
Your best bet is to KNOW about the robot, FIND it's most exposed sensors, and SMASH those with a crowbar or with bullets.
Bethesda, Md.: All our thinking is premised on the notions of sexual reproduction and of discrete organisms. To be sure, it is hard to tell when one coral begins and another one ends, but animals advanced enough to have emotional drives compete for mates and for -individual- survival. We do and will in fact make individual, free-standing robots, but most of the computational power will be distributed all over the place. So we can't kill off the enemy, because he and/or she will be everywhere and nowhere.
Daniel H. Wilson: What?
Wait, is this why those feral Sony AIBO dog robots that escaped into the Outback have been stealing babies in Australia?
Houston, Texas: Do you think Microsoft's leap into robot software will make the robots stronger or more vulnerable?
washingtonpost.com: Microsoft sees future in robots
Daniel H. Wilson: It will make the robots more mundane. In the future, when Microsoft leaves a security-flaw in their code it won't mean that somebody hacks your computer. It will mean that somebody takes control of your servant robot and it stands in your bedroom doorway sharpening a knife and watching you sleep. (Instead of just watching you sleep.) Creepy!
Blacksburg, Va.: Robots probably won't be nearly as good looking as Tricia Helfer (Caprica-6, the blonde one) from Battlestar Galactica, will they? I think an uprising would be much easier to accept if they were.
On a more serious note, robot designers in the home keep going down the humanoid path. As is evidenced by video game demonstrations for the PS3 and XBox360, the Uncanny Valley is getting deep, both for virtual humans, and for mechanical approximations.
How deep will the Valley turn out to be for robots? Do you think designers will try to make robots with less functionality but closer human approximations, or more functional robots that don't even resemble human body structure.
Daniel H. Wilson: The Uncanny Valley is a phenomenon in which people seem to get freaked out by robots that are *almost* human, but not quite. Nobody knows exactly why it happens, but a walking corpse is a good example of something that is almost human (looks like a person, walks) but not quite (not breathing, moaning for brains).
I personally don't believe the Uncanny Valley exists. It's just that people get freaked out by badly designed robots. People are also freaked out by ugly, scary people.
A lot of other researchers are also unfazed by the Uncanny Valley findings and so they are pushing the development of androids. Others aren't. We'll end up seeing all types of robots and the best designed ones will win.
I guess what I'm saying is that properly designed robots will be totally smoking hot. Woo!
Alexandria, Va.: Last night my wife said if I ever cheated on her she'd kill me so dead that I couldn't even come back as a zombie. Is that possible?
Daniel H. Wilson: I'm not a Zombie guy, so I couldn't tell you. However, I *can* say that if you cheat on your wife with a robot, it doesn't count. Unless you fall in love.
Reston, Va.: Wouldn't it be wise to put regulatory limits on robotics development in order to block anti-social consequences of robotic research and development?
Daniel H. Wilson: This is an absolutely terrible idea.
Robots are machines designed to help people. Fact: Most robots are designed to teach children, keep the elderly healthy, and protect puppies and kitties from innovation-stifling people like you.
If some company built a machine that malfunctioned and hurt people instead... well, they would either go out of business or have to start selling that machine to the military.
Plus, laws exist. So whether you kill somebody by unleashing your 500-ton atomic kill-bot or whether you personally stab them in the neck with a screwdriver -- you still goin' to jail.
Alexandria, Va.: How can I tell if my co-workers have been secretly replaced by androids? They seem much more productive today.
Daniel H. Wilson: Tap your co-worker lightly on the head with a crowbar. If there is a metallic ringing sound, run away. If there is not, run away twice as fast.
Baltimore, Md.: If in 25 years or so AI has greatly surpassed human intelligence, eventually thousands or millions of times smarted, as Kurzweil predicts, will there really be any defense against it, via robots, gaining total control? And will that be a bad thing for humanity.
Daniel H. Wilson: The whole idea of the singularity -- in which robots get smarter than humans and then suddenly go nuts and keep gaining intelligence at exponential rates -- seems silly to me.
Expecting this to happen by itself is like filling a box with legos, zapping it with lightning, shaking it really hard, and then expecting Mecha-Godzilla to pop out.
Research in robotics is iterative, it progresses step by step. It takes thought and time. If super-intelligent robots are coming our way, we'll know about it in plenty of time to give away our pets and start building bomb shelters.
Houston, Texas: Do you think robot toys are bad for kids?
Will they brainwash them?
Daniel H. Wilson: This is a cool question! There is a lot of research going on right now to find out how children interact with robot toys. So far, we know that in the short-term a child will treat a robot dog like a real dog (petting it, shoving bones in its mouth, etc,).
But what about the long-term? How does it affect a kid to grow up without ever having to feed the pet? Or the pet never biting the dust? Will that kid be emotionally stunted? What happens?
I don't know! But Sherry Turkle at M.I.T. does and so does Peter Kahn at the University of Washington. So Google 'em already!
Elkton, Md.: If robots do take over the world, will we be able to defeat them in simplistic ways like Vampires can be defeated in fiction? (ie Sunlight, chopping off the head, stake through the heart, etc)
Better yet what is the probability there will be a human savior who is the robot slayer? And then what are the chances that this robot slayer won't be just another robot itself? Or a robot slayer that not tied to any major corporation?
Daniel H. Wilson: Of course there will be a human savior to save us from the robots.
It will go something like this: a washed up alchoholic robot-fighting trucker who just wants to get his kid back will reluctantly take on the robots and ultimately win both the respect of his son and the truck of his dreams.
See, I didn't get a PhD for nothin'!
Sci Fi: When I think of robots I don't think of machines on an assembly line that do the work of humans. I think of 'The Terminator' or 'Six Million Dollar Man' episodes where he and the bionic woman had to fight robots who were actually stronger than bionics. It was so cool when the robots' faces would get knocked off and you could see all of the exposed circuitry where their eyes, nose, and mouth used to be (very scary for a 10 year old).
How soon will we see robots in real life that look human? I want one that looks like a super model.
Daniel H. Wilson: Don't we all.
Good news: Extremely human-like robots already exist. I suggest you check out the robots that David Hanson has created. David's robots are so lifelike that during experiments he routinely leaves off the back of their heads, so that test subjects know they are in the room with a robot.
Plus, if you like the thought of tongue kissing Albert Einstein (and who doesn't?) then you'll love David's creations. www.hansonrobotics.com
Morrisville, N.C.: When will robots have standing to sue and be sued in court? Will they need guardians ad litem?
Daniel H. Wilson: Not any time soon.
Robots haven't done anything deplorable quite yet, but just like any appliance, eventually one will malfunction and eviscerate a human. A well-designed robot will have a "black-box" so that it will be possible to step through its decision process and find out why it did what it did. It could be possible that the creator would liable for damages caused by a rampaging, badly designed robot.
An interesting parallel might be to look at air bags. They sense the environment (waiting for a crash), decide to inflate (upon crash), and then inflate. Usually they save people. Sometimes they hurt people. But we don't take them to court.
Tinseltown: I forgot if it was Leno or Letterman who asked this, but let me please pass along the question: if I make love with a robot, am I cheating on my blow up doll?
Daniel H. Wilson: Yes, you are. Scumbag. But you aren't cheating on your blow up sheep.
Washington, D.C.: Robots as personal servants:
How soon? Where can I find one? I'd really like 'someone' to clean the bathrooms, and grocery shop, and pick up the dry cleaning. Can I program him/her/it to also give me tips on the ponies after lots of analysis?
Daniel H. Wilson: Look to Japan. The Japanese have a rapidly growing elderly population and strong cultural push for the children to care for their aging parents. And you know what that means -- lots of robots!
Servant robots will be bathing elderly Japanese women before they ever see your dry cleaning.
Bethesda, Md.: Surely, the wars are far more likely to be between one species of robots and another, rather than between robots and humans. Our survival will be wholly incidental. Why hasn't anyone recognized this?
Daniel H. Wilson: Because we're all really, really stupid.
But seriously, the preferred nomenclature is "model" not "species" -- the robots aren't alive and don't evolve the same way as good old fashioned Earth life.
What you describe is essentially an arms race. One robot trumps another, so the next model runs on diesel fuel and shoots rockets out of its ass. The interesting thing about robots is that they can potentially design newer robots OR simply modify themselves in real-time.
It could prove to be a really fast arms race that culminates in the destruction of everyone who hasn't read my fabulous book. And that's a fact, smarty-pants.
Washington, D.C.: Mr Wilson,
Do you have a favorite fictional robot?
And do we have more to fear from Terminator style killer machines or the sneaky, subservient types like C-3PO?
Daniel H. Wilson: I do like the rattled, subservient types of robots. But I don't respect them.
Just to drop names: Yesterday I was hanging out with C-3PO. Actually, the guy who wears the suit and does the voice and plays C-3PO in the movies. He came to Carnegie Mellon University for the Robot Hall of Fame induction ceremony. And let me just say -- he is a really fun, nice guy. He can't take a punch though.
And back to the question. The robots we should fear most are the ones we love most -- the ones that are in our homes and workplaces.
Daniel H. Wilson: Okay, thanks for another round of fantastic questions. I wish that I had time to answer more of them seriously, but you can visit www.robotuprising.com for more crucial information on how to keep your blood inside your veins when the metal ones come for us.
HOW TO SURVIVE A ROBOT UPRISING is ready and waiting to save your life on Amazon and at B&N and Borders and everywhere else. Get it before the robots buy them all up.
Editor's Note: washingtonpost.com moderators retain editorial control over Live Online discussions and choose the most relevant questions for guests and hosts; guests and hosts can decline to answer questions. washingtonpost.com is not responsible for any content posted by third parties.