Robot ethics is always a pretty hairy subject to tackle, largely because human ethics is something that’s hard to reduce to logical rules, and robots really, really like logical rules. There are ways to try and make it work, though, and researchers at the University of Connecticut have been trying to combine machine learning with traditional ethical philosophy to teach robots to behave ethically.

The approach they’re taking is based on a technique pioneered by a philosopher named David Ross, who’s automatically awesome because he’s Scottish. Ross suggested that when people make ethical decisions, they’re actually balancing a bunch of different variables against each other, including things like ‘do good,’ ‘don’t do harm,’ ‘keep your promises,’ ‘don’t be annoying,’ and other things in that vein. Variables, you say? You know what are good with variables? Robots, that’s what.

In this demo, for example, Nao is trying to give medication to someone who doesn’t want it. The robot is considering several variables here, including the importance of taking the medication at all, the importance of taking the medication at a specific time, and the implications of the patient’s refusal. Essentially, Nao does some math to balance the values of these variables, and decides that the patient gets to refuse their meds once, but after that Nao appeals to a higher authority (the doctor, who we can probably assume is a human).

So where do the values for all of these variables come from in the first place? That’s the tricky bit, but for things like medications, you can see how it would be possible to come up with figures. Other values (like the importance of respecting the wishes of a human) are much harder to quantify, but part of the hope is that robots will be able to observe the results of their ethical decisions, and use that information to help them make better choices in the future.

[ UConn ]

Last August, we posted about some swarm robots at EPFL that evolved deceptive behaviors to fool other robots into starving to death (or something). These behaviors were passive, however, in that the robots weren’t making intelligent decisions to deceive other robots, but simply executing evolved behaviors where deception improved overall fitness.

Researchers at Georgia Tech have taught a robot to be able to identify situations where deception might be a good idea, and then use deceptive tactics to help it accomplish its goals. While the immediately obvious application is militaryish (and the research is sponsored by the Office of Naval Research), deception is potentially important in lots of other situations. One example is a search and rescue scenario, where a robot might be better off deceiving victims than informing them that (say) there’s a 93.77% chance that everything is about to go terribly horribly wrong.

Their first step was to teach the deceiving robot how to recognize a situation that warranted the use of deception. A situation had to satisfy two key conditions to warrant deception — there must be conflict between the deceiving robot and the seeker, and the deceiver must benefit from the deception. Once a situation was deemed to warrant deception, the robot carried out a deceptive act by providing a false communication to benefit itself. The technique developed by the Georgia Tech researchers based a robot’s deceptive action selection on its understanding of the individual robot it was attempting to deceive.

To test out the deception programming, the researchers had their robots play hide and seek. The hiding robot had a choice of three hiding spots, each with a set of markers in front of it. Recognizing that the situation could benefit from a deceptive tactic, the hiding robot would knock down markers in front of one hiding spot, and then go find a different spot, avoiding the markers. This technique fooled the seeking robot 100% of the time, as long as the hiding robot was able to knock down the correct markers.

One of the researchers involved in this project is Ron Arkin, whom you may remember from a few previous posts on robot ethics. And it’s no coincidence that he’s involved in sneaky robots:

While there may be advantages to creating robots with the capacity for deception, there are also ethical implications that need to be considered to ensure that these creations are consistent with the overall expectations and well-being of society, according to the researchers.

“We have been concerned from the very beginning with the ethical implications related to the creation of robots capable of deception and we understand that there are beneficial and deleterious aspects,” explained Arkin. “We strongly encourage discussion about the appropriateness of deceptive robots to determine what, if any, regulations or guidelines should constrain the development of these systems.”

Robots capable of deception is a pretty serious thing. Of course, it has to be understood that when we say robots being deceptive, we mean robots that we’ve programmed to execute behaviors that appear deceptive… It’s not that inherently, some robots are somehow deceptive. I suppose you could argue that there’s no functional difference, but the important thing to remember is that robots do what we tell them to do. If we tell them to be sneaky, they’ll be sneaky. And if there are potential ethical issues with any of this, look to the humans, not the robots.

[ Project Website ] VIA [ Physorg ] and [ GT News ]

We posted about three different telepresence robots yesterday: the Anybots QB, the Willow Garage Texai, and the Vgo. Telepresence is great in concept, but as Erico Guizzo discussed a bit, it’s a strange combination of being somewhere and not being somewhere, and interactions with people are different in ways that range from subtle to drastic. Willow Garage has been using Texai in their office for quite a while; one of their employees, Dallas Goecker, ‘commutes’ daily from Indiana to California via Texai. So, they’ve been figuring out some of these social rules as they go, to the point where some things are now a part of the Texai communication software:

Here are a few built-in bits of etiquette:

Texai Rule #1: If you see me, I see you. Explanation: It’s about two-way communication. Implications: The cameras face forward because the screen faces forward. The pilots are only allowed to drive the Texai once they’ve shared their video stream.

Texai Rule #2: Texai do not record audio or video. Explanation: It’s about face-to-face communication.

More, after the jump.

These are some generalized social rules that apply to the Texai:

* When a pilot wants to get a local’s attention, they’ll hover by the office window or open doorway (initiating conversations). The extreme example of this is running into the doorway to “knock.”

* When working from outside of the building, WGers will often sit their Texai at their real office desks because that’s where other people know to go find them.

* When Dallas goes down the hallway, people often say hi to him; this has happened less so now that Texai is not a novelty, but it happens at about the same rates as when we say hi to each other in the building in person.

* Dallas will turn the Texai and camera just as he turns his body and eyes during meetings to show where he’s paying attention (non-verbal communication)

* Pilots like Jonathan Knowles and Sergey Brin at the X-Prize were navigating the cocktail and dinner rooms much like a person would navigate a party, slowing down when approaching groups of people talking and going a bit faster when going between groups (respecting personal space)

* Locals will push chairs around the table to make room for the Texai at meeting, which is much like pulling out a chair for someone to sit down at the table. Texai are actually able to push many types of chairs themselves, but it’s just a nice thing people do for Texai pilots.

* Just like cell phone etiquette, one of the top questions people ask when encountering a new person in a Texai is, “Where are you?”

Of course, this is not to say that all telepresence robots are the same, or that all of these points of social etiquette will apply to every environment. But it’s interesting to hear about what actions and interactions are natural when a person is sort of both there and not there (in terms of physical presence), since it’s a rather unnatural (and perhaps unprecedented?) state in which to exist. It’s funny, too, how people instinctively associate the person controlling the robot with the robot itself: it becomes entirely natural to address the Texai as “you” as opposed to “your robot” or something like that.

I haven’t interacted with anyone through a QB or Vgo, but I would imagine that it would be significantly more difficult to identify a person as one of these robots due to the lack of a big screen with the pilot’s head displayed on it to allow for face to face identification and communication. And, this seems to be one of the things that make the Texai such effective telepresence tools: the ability for the pilots to be instinctively treated just like a person, in a robot.

Many thanks to Ken Conley and Leila Takayama of Willow Garage for talking to me about Texai.

[ Willow Garage Texai ]

When we first met the Kokoro I-FAIRY robot at CES in January, we were kinda wondering what it, like, you know, did. Apparently, it does weddings, of all things (actually, it’s a visitor guide, but that’s less interesting). The bride, Satoko Inoue, works at Kokoro, and her husband Tomohiro Shibata is one of their clients.

“It’s true that robots are what caused us to first begin going out, and as suggested by my wife, we decided that we wanted to try this sort of wedding,” Shibata said after making his vows.

After saying “I do,” the bride said that she wanted to use her wedding to show people that robots can easily fit into their daily lives.

“I always felt that robots would become more integrated into people’s everyday lives. This cute robot is part of my company, I decided that I would love to have it at my ceremony,” Inoue said.

That’s pretty cool, but is a robot actually allowed to conduct a marriage ceremony? Well, in the US, marriage is up to the individual states. Here in California, the law says the following:

Marriage may be solemnized by any of the following who is of the age of 18 years or older:

(a) A priest, minister, rabbi, or authorized person of any religious denomination.

That word “person” in there probably was not intended to specifically exclude robots, but until robots are people too, they’re probably not allowed to perform marriages, at least in California. In Illinois, on the other hand, it’s a bit different:

A couple can get married if the marriage is “solemnized” by a judge, a county clerk, or “in accordance with the prescriptions of any religious denomination, Indian Nation or Tribe or Native Group, provided that when such prescriptions require an officiant, the officiant be in good standing with his religious denomination, Indian Nation or Tribe or Native Group.”

So in Illinois, you just need an officiant, not a person. And if your religious denomination involves robots, you might be able to make that work. Really, it’s not so much that marriage by robot is specifically prohibited by law in some states, it seems to be more that the law has some catching up to do with technology.

[ Reuters ] and [ Internet Cases ] VIA [ Robotics And The Law ]

It is not hard to imagine why robots raise privacy concerns. Practically by definition, robots are equipped with the ability to sense, process, and record the world around them Robots can go places humans cannot go, see things humans cannot see. Robots are, first and foremost, a human instrument. And after industrial manufacturing, the principle use to which we’ve put that instrument has been surveillance.

…

The home robot in particular presents a novel opportunity for government, private litigants, and hackers to access information about the interior of a living space. Robots on the market today interact uncertainly with federal electronic privacy laws and, as at least one recent study has shown, several popular robot products are vulnerable to technological attacks—all the more dangerous in that they give hackers access to objects and rooms instead of folders and files.

Okay, so, should you panic? Probably not. More, after the jump.

As is already occurring in the digital space, roboticists might simultaneously begin to develop privacy enhancing robots that could help individuals to preserve their privacy in tomorrow’s complex world. These might include robots that shield the home or person from unwanted attention, robotic surrogates, or other innovations for now found only in science fiction.

This is interesting, especially in terms of parallels with the computer industry… The first computers inspired (I guess you could say) the first computer viruses, which in turn prompted the first pieces of antivirus software, and so on. Even before we hit the science fiction point, it’s likely that we’ll see things like anti-drone drones. On an individual basis, though, it might be realistic (soon, even) to see robots that exist purely to control your privacy. For example, you’d feed the robot your mail, and it would scan and shred it. Software side, the robot could spend its time monitoring your financial information, removing you from Google and Facebook, and de-spamming your email.

Of course, having such a robot would itself present a privacy concern, but it’s sort of like just using a computer (or, the internet) in general: by doing so, you’re making compromises to your privacy. There are ways to mitigate those compromises, but it’s the price you pay for access to technology. There’s no way to exercise complete control over information that you make available to others, and if consumers decide to embrace that aspect (i.e. telepresence) of robotics (which is one of the most appealing aspects), they need to understand that there is a potential for abuse the same way that there’s a potential for abuse whenever you make a phone call, or use your credit card over the internet. There’s no need to panic, but there is need to be aware.

I highly recommend reading the entire chapter… There’s a lot of thought-provoking type stuff, but it’s obvious that as far as actual privacy law goes, technology is outpacing relevant legal structure. Unfortunately, what seems likely to end up happening is that what privacy means in terms of robots will be decided as more of a reactive damage-control measure rather than something proactively implemented. Oh well… That’s what we get for not slowing down to let society catch up, I guess.

[ Robots And Privacy ]

Thanks Ryan!

Kevin Kelly from Wired recently interviewed Jamie Hyneman and Adam Savage from Mythbusters for the Commonwealth Club of California. As part of the interview, Adam and Jamie were asked (somewhat jokingly) whether they’re afraid that machines will take over in the future, particularly with regards to the present development of armed robots.

Now, if you’ve been reading BotJunkie for a while, you’re probably aware that this is one of my favorite subjects to harp on, and as much as I respect admire worship ::cough:: like-in-a-strictly-professional-manner Adam Savage, I won’t let you down.

So, is it a bad idea to give a machine a gun? Of course it is. It’s a terrible idea. But guns were a terrible idea to begin with (from a lofty ethical viewpoint, anyway). The terrible part about guns is that guns can kill people, and not giving guns to robots isn’t going to change that. Really, the question should be, is giving guns to machines a better or worse idea than not giving guns to machines? This where I think armed robots have a use.

I guess fundamentally, the part that I don’t understand is where Adam says that he knows how machines work and he wouldn’t trust a machine with a gun. I wonder, though, if we have a better idea of how most machines work than how some humans work… Like, it’s a fundamental right for humans to have guns, and there are plenty of humans out there who are far less predictable or reliable than a robot. It’s certainly true that robots are more prone to things like mechanical failures, but we already entrust our lives to robots on a daily basis (often without realizing it). Giving a robot a gun is just an especially obvious way of making it dangerous.

I hate to keep coming back to this analogy, but it’s like driving a modern car: between things like anti-lock brakes and cruise control and (now) parking and lane assist features, your car (if you have a fancy one) has the ability to control your brakes, your accelerator, and your steering. If you have power windows and door locks, it has control over those things, too. It’s only designed to be autonomous in very specific situations, but what we’re talking about here is mechanical (or software) failure. And generally, that just doesn’t happen, because cars have been designed and tested with safety and reliability in mind. I don’t see why it couldn’t be any different with armed semi-autonomous (or even autonomous) robots.

Whether or not it’s ethical to arm robots is (I’d like to think) a separate issue. The short answer? No, it’s not. As has been pointed out, arming a robot makes it easier to resort to violence since the consequences are much less severe. Some people might even argue that that reason alone should keep robots out of combat, but I doubt that those people are going into combat themselves. My guess (and this is really just a guess since I’m in no way qualified to make any other sort of comment) is that if some kind of armed conflict is inevitable and there is substantial risk of injury or death, most people who’d be directly involved in that conflict would rather send a robot in their place if possible. And, that’s really what it’s all about: robots doing the dangerous things so that humans don’t have to.

[ FORA.tv ] VIA [ Gizmodo ]

As much as I’d like to, I can’t really let this one pass… 31 years ago today (January 25th), Robert Williams was killed by a robot arm as it was retrieving parts from a storage facility at a Ford Motor plant. It was the first recorded death of a human by a robot. Williams’ family was awarded $10 million in damages due to lack of safety measures surrounding the robot… This was largely the same situation as the accident that happened more recently in Sweden.

[Ronald Arkin] described Williams’ death as an “industrial accident,” one in which the lack of physical safeguards were at fault. The death was not caused by the robot’s will, he cautioned.

“It was not an ethical lapse, unless you’re a Luddite against the Industrial Revolution,” Arkin said in a recent telephone interview.

Remember this, because this sort of thing is going to happen in the future, and inevitably it’s going to be outside of an industrial setting. Whatever you may think, and however it may look, and whether or not any particular robot can be anthropomorphized, it’s not willful. It’s not about ethics. Robots are machines, people program robots, and accidents happen.

VIA [ Wired ]