About Me

My photo
I have a burning need to know stuff and I love asking awkward questions.

Saturday, November 10, 2007

Forget robot rights, experts say, use them for public safety

The Guardian

Tuesday April 24 2007

Scientists have criticised a government report which advocated a debate on granting rights to super-intelligent robots in the future as "a distraction". They say the public should instead be consulted over the use of robots by the military and police, as carers for the elderly and as sex toys.

The robotics experts were commenting on a report published by the Office of Science and Innovation's Horizon Scanning Centre in December. The authors of Robo-rights: Utopian dream or rise of the machines? wrote: "If artificial intelligence is achieved and widely deployed (or if they can reproduce and improve themselves) calls may be made for human rights to be extended to robots."

The idea of robots becoming so smart that they acquire a conscious sense of self has fascinated science fiction writers for generations. The recent films I, Robot and Bicentennial Man, both based on books by Isaac Asimov, dealt with the question of whether intelligent robots should enjoy human rights. In the first a policeman played by Will Smith tracks a robot called Sonny that has apparently gone against its programming to commit murder. In the second Andrew the robot embarks on the quest for equal rights. But the scientists said true robot intelligence is so far in the future that it should not be treated as anything more than science fiction. "It's really premature I think to discuss robot rights," said Owen Holland, a computer scientist and expert on machine consciousness at Essex University. "[This report] is certainly not based on science and it is not realistic."

Noel Sharkey, a roboticist at the University of Sheffield who is a regular contributor to the BBC's Robot Wars, agreed, but he said there were more immediate concerns. "The idea of machine consciousness and rights is ... a bit of a fairy tale as far as I'm concerned," he said. "My concern is about public safety. I think we need proper, informed, public debate about where we are going with robotics at the moment. We need to tell the public about what's going on in robotics and ask them what they want." Last year the South Korean military unveiled a robot border guard built by Samsung that can shoot targets up to 500 metres away. He said these could be programmed with a shoot-to-kill policy. The US, meanwhile, is on the way to achieving its goal of replacing one third of its ground vehicles with autonomous robots.

"It would be great if all the military were robots and they could fight each other, but that's not going to be the case," he said. "My biggest concern there is that it goes against the body bag politics. If you don't have body bags coming home, you can start a war much more easily." Once robots become more common in warfare, he predicted they would be used more widely in policing and surveillance; so far there has been very little serious and informed public debate on these issues. Offenders could, he suggested, be monitored at home by a guard robot and the streets could be patrolled by mobile robot CCTV. They could also be used to deal with riots and other civil disturbances, he predicted. "Imagine the miners' strike with robots armed with water cannon." By providing companionship and basic care and health monitoring for older people, robo-carers could look after the increasing numbers of elderly people. And he predicted that vibrating sex-robots would be available soon for those bored with blow-up dolls.

[Although it brings up some interesting ethical issues, the idea that we should be thinking of ‘robot rights’ is highly premature. We are, in my opinion, decades (at least) away from any kind of AI – Artificial Intelligence – comparable with our own. As some of the comments above point out what is of much more immediate interest is how the ever expanding use of robots in all walks of life needs to be monitored and controlled if necessary. We are already seeing the use of semi-autonomous robots on the battlefield and this can only increase. I don’t think that it’s too far fetched to see fully autonomous fighting machines before mid-century. Once that happens it won’t be long before robot police are patrolling our streets. It’s been a long time coming – accompanied by much undeserved hype – but it’s likely that within our children’s lifetimes robots will become a common sight on our streets. What we need to think about is what exactly we want them to be doing and how we go about restricting what they can and cannot do. Any discussion of Rights comes much later.]

9 comments:

Karlo said...

I agree with concerns about weaponization. When robots can patrol streets, a whole different police state becomes possible in which the elite rulers don't even bother with trying to gain public acceptance. So I'm more concerned with the new pentagon robots than I am about box-cutters.

CyberKitten said...

I understand that some police forces are already using UAVs (Unmanned Aerial Vehicles) to patrol over inner city areas. How long before we see ED209 on the streets [grin].

Scott said...

Silly robotic experts, governments can't grant rights.

CyberKitten said...

scott said: Silly robotic experts, governments can't grant rights.

[grin]. I think we've been here before....

Juggling Mother said...

"governments can't grant rights.
"


They can take them away. Scott has on occasion moaned about the gov't restriction, curtailing or removing his right to......

If they can take them away they can give them back.

Or just give them.

Although I thibnk we're a fair way off worrying about robot rightsn yet - Asimov was a little out on his timings:-)

CyberKitten said...

JM said: Although I think we're a fair way off worrying about robot rights yet - Asimov was a little out on his timings:-)

I doubt very much if we'll see decent level AI in less than 30-50 years. Though I guess if there's a serious breakthrough..... However, I don't expect to read about a Robot Rights movement in my life time... in our Grandchildrens maybe... [grin]

Scott said...

They can't take them away or give them they can only infringe them or respect them. Rights, by definition, cannot be give or they are not rights at all, rather privileges. This is, as Jefferson noted, self evident.

Robots, for instance, can have no rights to speak of because they are created by humans and therefore depend on humans for all permissions. The only privileges they have are those given to them by their programmer.

The only way a person can talk about robots being given rights is with an incorrect definition of "rights".

sirkolgate said...

I think the rights we need to worry about are the people being either attacked, policed, or cared for by the robots. I agree with the Owen Holland (the computer scientist and expert on machine consciousness) in that there is too large of a gap between our ‘reactive’ robots of today and the ‘conscious’ robots of the future.

CK I’d place my true AI estimates at hundreds of years even with the advancement of technology that we see today, not mere decades. I write programs and as sophisticated as the logic has become and the remarkable advancements in ‘sensor’ technology there is no perceptive bridge between a ‘responsive’ robot and a ‘conscious’ one. The ‘brain’ of a robot is code, and that code may allow for a database to be compiled which gives it a better ‘mathematical’ understanding of its environment but there is nothing in today’s ‘computer brain’ that lends itself to ‘behavior’. You just simply can’t write code that performs outside its specifications. Well you could throw in some code driven by random number generators but that’s chaos, not intelligence.

That’s not to say that you can’t write code that doesn’t do what you want. Every programmer knows that programs often seem to have a ‘mind of their own’, but even the most ‘mysterious’ of errors is always tracked down to a flaw in the logic or hardware.

Also, I work with large robotic machines in my job (saws and CNC routers that cut specified designs into wood doors and panels) and they are top of the line, million dollar machines that are still prone to frequent breakdown and programmatic bugs.

So, my expectations for robots that are perceptive enough to police the streets or care for the elderly seeing release within our lifetimes is slim to nil. As a programmer I try to imagine being shouldered with the task of defining the ‘ethical rules of street engagement’ into a program, as well as all the laws and subsequent changes to civil law ever year, it’s a nightmare.

CyberKitten said...

Thanks for that sirkolgate. Highly informative.

sirkolgate said: CK I’d place my true AI estimates at hundreds of years even with the advancement of technology that we see today, not mere decades.

Oh, I wouldn't put it at *hundreds* of years.

It is of course possible that AI is simply beyond us - or beyond artificiality. Intelligence might have an intimate link with biological 'stuff'. There might be some reason we have yet to discover which prevents machines from achieving intelligence. I'm not convinced by this - but it is possible.

On the other hand the difficulties we are having in the production of AI may have something to do with the fact that we don't really understand natural intlligence yet. Without such an understanding it might be more difficult or even impossible to create AI.

It's certainly conceivable that there might be a breakthrough of some kind that allows us to produce true AI but I couldn't even guess at how far away we are from that. I certainly don't think that it's just a case of building a sophisticated enough database of rules and a fast enough search engine to run them. Rule based AI will almost inevitably fail when it meet new challenges - which it will do on a regular basis.

The only direct experience of AI I have is in computer games. I find that either its scarily good or really stupid - and often a mixture of the two. Some things it does *really* well whilst other often very simple things it just can't cope with. I've heard that various AI's are going to be trialed in on-line environments which will allow them to learn (and fail) with safety. I think I've already met some of them in on-line battles [grin].

sirkolgate said: As a programmer I try to imagine being shouldered with the task of defining the ‘ethical rules of street engagement’ into a program, as well as all the laws and subsequent changes to civil law ever year, it’s a nightmare.

It could be a huge problem but I think that attempting to create a viable and flexible set of 'rules of engagement' is doomed from the start. AI's will probably have to be 'grown' and 'evolve' rather than be written from the ground up - but my knowledge of the whole area is..... limited.... so what do I know [grin].