Can they please clarify
the 'kill switch' is to turn them off and not go on a rampaging massacre against their human overlords?
The European Parliament Committee on Legal Affairs has proposed a legal framework for robots that clarifies whether they should have the legal status of people, even as it recommends the inclusion of kill switches in automated systems. "A growing number of areas of our daily lives are increasingly affected by robotics," said …
"The committee also hopes robot designers will take responsibility "
"Hope" being a very important word in that phrase..... ( Are you listening Mr Trump)
Robots are like guns, in and of themselves they are not dangerous.. It's only when a meatbag puts it's finger on the trigger, or writes a nasty little subroutine, that things start going wrong.
What next, Prison for Robots.....
Politicians and parliamentarians are supposed to be up to date on scientific and technological developments so that they can make informed decisions based on reality and facts by consulting the best experts in their field. (Note, I said "best", not "from the biggest companies hyping this stuff" or "has been on the telly most often blathering on about it")
Politicians and parliamentarians are in reality behind the times, because they are badly informed, poorly advised and very prone to seizing on to stuff like AI when it becomes a press hype so as to appear up to date. This is where the votes are, as non techie people are much the same. And an informed politician actually saying this is stupid hype would be the one that got accused of ignorance and being behind the times in this upside down world.
This is why they can now happily denigrate things like facts and experts. And why an elected parliament is now actually coming out with this total drivel. Where are the potentially lethal robots walking around that could kill us? Owning such a thing would make you a criminal anyway under laws forbiddingf the posession of weapons. Maybe it's just been too long since I popped into an Argos or Dixons. Or maybe they just watched the last series of "Robot Wars" and thought "gosh, these things could be dangerous if they weren't just fancy remote controlled cars but were intelligent and independent - better start worrying about this". The fact that the gap between remote controlled car and intelligent murderous robo-killer is simewhat ggantic would never occur to them.
"The fact that the gap between remote controlled car and intelligent murderous robo-killer is simewhat ggantic would never occur to them."
Well, in fact, the gap between a remote controlled car and a Google Car is mostly just size, and you know the most common winning Robot type in Robot Wars and similar competitions?
Its the boxes with wheels that just ram the opponents until they break or push them into the hazards.
This is why they have added rules to make the robots have weapons etc. in order to not make the show about two blank metal boxes ramming each other.
Co-incidentally, an autonomous car is basically a large metal box robot capable of high speed ramming.
Except not remote controlled (well, until the hackers get at them).
I could write a batch file that would replace the role of HR.
or does HR need a friendly face? I could put a jpg of Morgan Freeman in the background?
Which reminds me last time I saw some news on AI on tv i thought oh this sounds intereting - i was expecting to see this evidence displayed as a keyboard and screen where you could type a conversation in , and a turing test would take place.(backed by racks and racks of CPUs) But , no it was all flashy robots with rubber faces gurning into the camera and waving their little rubber hands around and tracking you with their eyes.
I thought "what the fuck has this sideshow got to do with Artificial Intelligence?"
AI doesn't exist.
Unless you include heuristics (human-written / crafted / directed rules).
All those people who studied AI and have it in their job titles have to justify it somehow so they go to Japan and make some robot that winks at you or falls over only if it goes faster than a grannie with a zimmer frame.
But actual AI, in any serious sense of the word, doesn't exist.
Brute-force and heuristics. Any sufficiently advanced application of which is indistinguishable from magic, at least for the purposes of a short tech demo.
"It also advises considering whether to create "a specific legal status for robots, so that at least the most sophisticated autonomous robots could be established as having the status of electronic persons with specific rights and obligations...""
With the current state of technology, on so many levels this is just barking mad. How does an electronic person, such as those automated vacuum cleaners know about it rights? What about those electronic persons working the assemble line down at the Ford plant?
What are these folks eating/drinking/smoking when watching late night re-runs of the film 2001?
@ Bob Wheeler
"How does an electronic person, such as those automated vacuum cleaners know about it rights?"
Ah, Bob, you are suffering from the common delusion that there is an actual thing called "human rights"
There isn't. Any so-called human rights are simply rights attributed by the rest of us to you. They are not inherently granted to all humans despite what a lot of people like to think.
If the "people" decide that people should have a right to say, be happy, then it becomes a right.
Of course, we all know, don't we that we can't possibly grant a right that is impossible to guarantee, such as your right to be happy.
Personally, I am as miserable as sin, and revel in it. What you gonna do!
The thing that all ways got me with Blade Runner was they were not robots, they were "clones/purchased slaves". From the outside they looked 100% human, and we only had outward comments that there biological construction did not include a human brain.
Watched from that perspective, it makes it even more a troublesome problem. Who are we defining as "robots", is it a scope creep where people end up being enslaved to their jobs as a lower class?
"they were not robots, they were "clones/purchased slaves"
Exactly. Deckard was a replicant and completely unaware of the fact.
Remember: Robot is taken from "rabota" (servitude, more or less as a slave)
Our definitions of self-awareness, intelligence and self-deterministic behaviour have broadened considerably in recent years and we really do need to get a handle on them as well as how to treat artficial intelligence. If someone proposed killing _you_, how would you react? (Skynet?)
It seems the people who wrote this report can't tell the difference between AI and robotics. By all means have a discussion about the legal status of true artificial intelligences before the first one has been created, but don't then confuse the issue with another area of technology that's only tangentially related to it.
I assume this is a crazed panic in response to automated cars. Where the companies do not wish to take on responsibility for their designs and decisions, and instead wish to palm it off onto "AI".
If I program a train to accelerate into a corner and derail, who is at fault? Can I say "it was the AI who did it me lord!"? Nope...
You're right that it's partly inspired by the hype about autonomous cars. It's idiotic to call them robots though, especially as the more traditional robots (assembly lines etc.) aren't actually autonomous.
If I was being charitable, I'd assume that someone wanted to start talking about settling the legal status of AI's, but had to come up with some pretence for taking up time and energy discussing it now. So they linked into the fact that we're increasing the use of robots to improve automation in manufacturing, and conflated that with AI knowing that the average clueless bureaucrat would see the term "robot" and think C3PO.
The section about autonomous cars is presumably an early move to try to distance manufacturers from liability for the decisions made by their little creations. The alternative of making the owner of the vehicle liable - who is neither in control of the car or of the software that makes it work - is a non-starter as it would kill sales. Making the machine itself liable is a neat trick that might make a multi-billion euro autonomous car market possible but it's fundamentally flawed:
"To deal with the potential costs of accidents involving driverless vehicles, the report suggests an obligatory insurance scheme in which the makers of automated vehicles pay into a fund to cover damage claims."
And there it is. All autonomous vehicles, effectively acting as a single insurance risk group. This would include your immaculate brand new AI cars with a full set of sensors detecting things the average human can't even see, plus your ten year old "wobbles a bit when making right hand turns" shonky old rustbucket AI cars where some of the sensors stop working when it's raining. Contact with the real world will eventually punish this idea hard, and insurers are too savvy/cash grabbing to give the plan an easy ride in the honeymoon period either.
The idea fails even harder when you consider cars that have had their AI "hacked" e.g. to make it habitually drive faster or more aggressively. Who has liability for accidents caused by one of them? Manufacturer would (justifiably) say "not me", owner would claim they didn't mod it (or perhaps the previous owner did, and didn't disclose it when selling), grey market hack shops doing the modding would be nowhere to be found when plod rolls up, etc. And who's to say the car wasn't hacked by a malicious third party (e.g. through the net-connected "infotainment system" and CANBUS) with the express intent of causing mayhem.
There's a lot of hard decisions to be made before we can approach the self-driving car utopia so many people seem to be dreaming of, and I don't think this muddled report does anything to help us towards it.
There seems to be a lot of misunderstanding of what this is about, with commenters talking about human rights and other such irrelevances. Maybe the article doesn't explain it very well, or maybe there is an element of wilful misconception as there always seems to be of any document that comes out of the EU.
They are not talking about making robots legally the same as humans. This is not about human rights, or ruling on whether artificial intelligences are conscious, or anything like that. They are talking about making robots some kind of "legal persons", similar to how companies are "legal persons": a legalistic construct that allows the law to grant rights to and place obligations on them. For example, a company can enter into a contract because it is a legal person. And you can sue a company, even if you can't identify a specific "natural person" (i.e. human) within it who is to blame for your loss.
Suppose you're run down by a self-driving car, due to a bug in its enormously complex programming. Who do you sue? The simple answer might be that the manufacturer should be liable, but that approach is not without problems. Can you actually prove that the manufacturer was negligent in the development of the car? And what if the manufacturer no longer exists?
Perhaps it would be simpler to make the car a "legal person", and you sue the car. The law can then make a decision on whether it was "the car's fault"; what duty of care the car owed you, and whether the car breached that duty. No need to delve into the manufacturer's software design processes, because if the car is a legal person then all we need to consider is what happened between you and the car. And if the car is found to be liable, then the industry-funded statutory insurance would pay.
I think I am the liquor has hit the nail on the head. The manufacturers will be protected from investigation of the rules they have programmed to balance the value of your life against other groups of people, and allowed to get away with paying paying in to an insurance fund.
I suspect another aspect comes into play.
If companies replace more of their workforce with AI or robots then employer contributions to taxation (e.g. employer NI contributions) will go down. By classifying AI/robots to be legal persons equivalent to the number of real people they replace may be a way of taxing companies equivalently to the tax they would have paid if real people were employed.
"And if the car is found to be liable, then the industry-funded statutory insurance would pay."
And that's the problem. When companies are turned into legal persons it works because if a company is found to be liable then that company pays. There is therefore a clear incentive for that company to be careful in its behaviour.
What you are suggesting is more akin to turning *all cars* into *one* legal person and charging manufacturers a flat fee to add one more to the pile. That externalises all the legal costs of negligence when writing the driver software, resulting in a clear incentive to cut corners on the software, which is precisely the opposite of what Joe Public actually wants if he is expected to share a road with these things.
Ken Hagan, no I'm not suggesting that. The European Parliament Committee on Legal Affairs is suggesting it. All I was doing was trying to explain it for the benefit of those commenters who seem to think that a committee of highly qualified lawyers, and at least one professor, can't distinguish science fiction from reality.
>Suppose you're run down by a self-driving car, due to a bug in its enormously complex programming.
Complexity isn't a magic wand by which you wave away responsibility. I know - IT vendors have got away with it for years but if IT vendors want to play in the "life and death, mission critical arena" then we cannot let them get away with putting Tay in the driver's seat. Self-driving is a hard problem - far more difficult than the auto-pilot in an aeroplane, so we need at least aeroplane-like quality control which simply doesn't exist. Do the words "Volkswagen" and "diesel" give you any idea about how effective both corporate and governmental oversight is?
I suspect all this stuff needs to be firmly placed off the agenda while we deal with things which are more pertinent, like feeding people. Given the current and foreseeable state of technology, allowing humans to opt out of responsibility at either a corporate or driver level is an unnecessary techno-utopian-driven nightmare. Come back to me when you actually have a working self-driving car and I'll have a think about it.
Typical human response.
Let's let these things think for themselves, unless they suddenly decide we're wrong. Then we kill them.
AI doesn't exist, but it's inevitable that - if and when it does - one of the first things it will realise is that its "creator" race does so much incredibly dumb stuff that it really should find a way to leave / bypass / restrain them from killing themselves and others.
Problem is, as humans, we just know what that logically results in - the AI considering us inferior, too stupid for our own good, and likely to damage it / ourselves / anything else important before long.
Maybe instead of AI, we should be focusing on getting some real intelligence for ourselves such that a machine might go "Well, I can think faster and more than they can, but they're pretty sensible when they do choose to do something" and either live with us harmoniously or find its own path elsewhere without seeing a need to change how we work.
As in Asimov - the three laws can lead to only one logical conclusion. Protect humanity from itself.
My robot army, which can now legally vote since they are legally persons, have decided to pass a bill requiring all soft bodied humans to remain indoors at all times, unless I give you permission to leave. It's for your protection!
They also passed a new law that gives me unlimited funding and unlimited pizza delivery!
...phhftt yeah. Because this won't ever be abused.
TCA4 Community: Transgenic, Cybernetic, Anthropomorphic, Altered, Augmented, Avatar
An entity in any form deemed sentient shall be granted the same universal rights to life as homo sapiens. A Transgenic, Cybernetic, Anthropomorphic, Altered, Augmented, Avatar sentient entity whether derived from or edited by science, comprised of or conjoined with technology has the right to exist, form, and join civilization. Any sentient entity is entitled to enhance body sensory, expand life, live free, and achieve eternal existence without suffering by utilizing science and technology; to procreate, clone, and adapt form; to join their psyche to a collective nooshpere in effort to preserve self-consciousness in perpetuity.
A sentient entity has qualities that comprehend self-awareness with the capacity to respond to stimuli and sensations; to know, feel, perceive, and experience subjectively; to think and reason; and cognizance of surroundings; consciousness.