Royal Academy Of Engineering Report On Autonomous Systems

Writing by Evan Ackerman on Tuesday, 1 of September , 2009 at 12:01 am

It’s been nearly a month since I’ve had a good rant about public perception of autonomous robots, so I was so excited to see an article on Economist.com entitled, “How real is the threat of autonomous technology?” The article is about a report released by Britain’s Royal Academy of Engineering on the social, legal, and ethical issues of autonomous systems, which you can read in full in PDF format here.

Commentary, after the jump.

As you might expect, the report is full of rhetorical questions and statements like, “debate and engagement on this issue would be of great value.” In fact, it’s not so much a report as, “here’s a bunch of questions that we have no idea how to answer but we’d feel better if smart people started thinking about them.” As such, there’s a mild undercurrent of something in the report that I suppose I’d better just call concern, as opposed to something less rational. But I can’t help feeling like these people are expecting the introduction of robots to be some kind of light switch, a major paradigm shift, as opposed to a gradual introduction that has already implemented itself in a whole host of ways that they’re not really appreciating the significance of precisely because their significance is not generally thought of as fundamentally significant. Um, if you read that last bit slowly a few times, I think it makes sense.

For example, let’s talk about robot cars, since there’s a whole section in the report about them, and you’re probably driving one. Most new-ish cars have already reached the point where the car, on some level, makes decisions for you based on what it decides your best interests are. Specifically, I’m talking about things like anti-lock brakes. ABS can brake a car better than you can on most road surfaces by avoiding wheel lock. This means releasing brake pressure. So basically, your car has the ability to not brake when you’re telling it to brake. Hypothetically, this means that a fault in the ABS system could keep you from stopping your car.

So why aren’t we panicking? Because that type of thing doesn’t happen often enough for it to be an issue. ABS systems undergo lots of testing, they’re redundant, and if something isn’t working, they disable themselves and tell you. There’s no reason these safety steps should not be the case with any other autonomous system, and as far as I’ve seen, they already are, in most cases. The point is that we (humans) are already making a compromise to our safety by giving up some control to semi-autonomous or autonomous systems that are smarter, better, and faster than we are. Overall, it provably makes things safer. Yes, it does add a small amount of risk that wasn’t there before, but the fact is, there’s always inherent risk in driving a car. There’s inherent risk (even if it’s just losing money on your energy bill) in relying on your autonomous robotic thermostat to keep your house a reasonable temperature. You have the option to decide not to use autonomous systems, if that’s something that concerns you, but most people know that in many important cases, robots are more reliable and/or effective at doing some very important things than we are. This same logic can be applied to other areas of robotics, including elder care robots and surgical robots.

One point that the report seems to consider particularly relevant is who we’d get to blame if a robot screws something up. I don’t like that way of thinking, but I do recognize that it’s probably going to come up at some point, so I think that debate and engagement on this issue would be of great value.

Just kidding.

The nice thing about robots, as opposed to humans, is that hypothetically (and I’d like to stress that hypothetically), decisions that robots make can ultimately be traced back to some instruction or combination of instructions that the robot learned (based on other instructions) or was given. Now, I’m not necessarily advocating that programmers should be responsible for robots having accidents, but it does potentially provide a way to distinguish between negligence on the part of a robot manufacturer (forgetting a program for an obvious contingency, for example), and an accident (or whatever you want to call it) on the part of the robot. The fact is, accidents happen. Car accidents happen all the time. Sometimes, they’re the fault of the driver or some other driver. Occasionally, they’re the fault of the manufacturer. And some of the time (and you could certainly argue even that), it’s just a combination of unfortunate circumstances. Even if autonomous cars are able to significantly mitigate accidents previously caused by human error, there are still (probably) going to be accidents. And to restate the point that I made earlier, driving a car implies an acceptance of risk. Every time to get into a car, you’re probably doing the riskiest thing of your entire day. If you choose to use a semi-autonomous or autonomous system to make your driving safer, which it will, you’re still implicitly accepting that there is going to be some risk involved.

I could go on in this same vein, but I won’t. I would like to say that the report does ask some potentially relevant, albeit slightly overblown, questions about how (say) autonomous and human controlled cars might interact in the future, but again, it’s the sort of question that I’m pretty sure is just going to slowly and calmly answer itself as the technology evolves. The overall point I’d like to make, I guess, is that it’s fine to ask breathless questions about all the complicated and ethically chewy issues that could arise when autonomous systems become a part of our lives, but we’re kinda past that already. Autonomous systems are everywhere, we just don’t think about them as autonomous systems, and that’s what makes me think that this transition that’s supposed to happen is already happening and it’s not going to be the problem that a lot of people seem to think it’s going to be.

[ Autonomous Systems Report (*.PDF) ] VIA [ The Economist ]

Comments (4)

Category: Artificial Intelligence

• Add this post to
• Del.icio.us -
• Digg

Name (required)

Mail (will not be published) (required)

Website

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>