Since such questions can potentially deepen our understanding of morality and ethics as well as intelligence and freedom of choice, it is never too early to pose them. A strat-up question could be whether there is a type or level of intelligence that is exclusively reserved for human beings.

These questions are not limited to synthetic intelligent systems. They have been part of our never-ending endeavors to realize our own very existence, from Hindu's eternal recursion and Buddhism ego-less interpretation of consciousness, to Descartes dichotomy of the soul and the matter that constitutes our physical embodiment.

Whereas some of the brilliant minds of our century such as Roger Penrose strongly believe consciousness is far beyond the grasp of artificially designed machinery, others such as Daniel Dennett find the idea not only plausible but also appealing and subject to further investigation.

Furthermore, the advancement of the topics such as Evolutionary Biology, Genetics, and Cognitive Sciences, whether Computational or those pertaining to living organism, has provided us with a better understanding of the processes that eventually lead to the emanation of sentient life forms.

I think it is reasonable to put our findings into practice, thereby taking the lessen from the world about us and leaving the myths aside: don't we misconduct? has the blind-folded tolerance provided any solution to the ignorance that our society has been desperately struggling with for centuries? and more importantly, isn't death a natural way of making room for the evolution of the species?

From my point of view it would be unethical not engage the problem of building a machine that has at least the simplest abilities of a human being. All around the world, workers are put to work in unsafe and unhealthy, physically and mentally unhealthy, conditions because human beings are the only known individuals that have the minimum cognitive abilities to perform these jobs.

As I see it, human kind has two options, whether it renounces to the products and comfort provided by these jobs, or it deeply commits to research for tools that will improve those workers living conditions.

This is an excellent question many of us thinking about or should think about.

My answer is simple but obviously keep the question rather open: I think that we can not and should not stop the research and the technological progress. We must think about the ethical consequences and make regulations to control the usage of any new technology as soon as possible. Therefore the pointers to blogs about legal philosophical and technological speculations regarding this issue are most welcome.

It is possible to design machines that can't be copied exactly.

Such is done against plagiarism. Suppose a robot with a

neural net that can't be copied (at least with our current

technology) is the only witness of historical talks, events,

or is the only one who has seen an antique papyrus scroll

before it crumbled into dust.

Such a robot would become part of our cultural heritage.

No one would dare to destroy him because of his value.

The value of this data.

Could we switch him off? The robot could threaten

to destroy the valuable data by himself if we don't

cooperate with him.

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2044797

http://news.bbc.co.uk/2/hi/technology/6200005.stm

http://blogs.computerworld.com/personal-technology/20938/if-apple-makes-robots-will-robots-have-rights

http://www.logos-verlag.de/cgi-bin/buch/isbn/1902

http://www.spiegel.de/wissenschaft/mensch/strafrecht-fuer-roboter-maschinen-sollen-fuer-eigene-fehler-haften-a-569004.html

http://www.heise.de/tp/artikel/15/15995/1.html

Regards,

Joachim

Thank you very much for the insightful and inspiring comments to my question. The motivation for asking it is that I am currently in the process of establishing a research group that aims at systematically investigating the extent to which human cognitive processes, in particular those related to situation awareness, may be applied to service robots. As I try to assess future consequences of our (and related) work, I cannot help but wonder: are we working towards a future in which robots enrich our everyday lives or are we looking at the modern face of slavery? Are we as researchers ethically obliged to limit ourselves in our ambitions to create artificial intelligent life?

There is no need to be emotional about machines with cognitive capabilities any more than we should be emotional about machines with anthropomorphic features. There is simply no equivalence between cognitive functions and sentiency.

It's not too early to think about this, for sure after some years it will be a strong debate subject. Not the level of intelligence achieved is the problem itself, but where and how the intelligence is used or will be used and how much harm can produce, either for a short or a long period of time.

Laura, you bring up a good point. I believe this discussion should be focused more on the ethical application of cognitive automation as it impacts the lives of humans as opposed to a discussion of rights perceived as inherent to cognitive systems.

It depends a lot on your definition of as well as your philosophical stance towards intelligence and cognition. If you believe, like John Searle, that a computer can never achieve consciousness and will always remain a mindless machine (cf. his Chinese Room Argument), then ethical considerations might not apply at all. But even if that were the case, an artificial intelligence that acts as if it is conscious, acts as if it feels pain or has emotional states, will prompt people to act in a certain manner to it. As such, people might decide to treat robots ethically whether it "makes sense" or not.

It is definitely not too early to think about the moral and legal status of robots. One of the first to be killed by a robot was Kenji Urada in 1981. While this particular incident happened due to the robot not being completely switched off during maintenance, there is always the question of whose fault it is, the programmer, the operator, or the robot itself. Note that this is not inherent to artificial intelligences and is a question that people wrestle with in aircraft crashes and the likes as well. Currently, we only seem to be satisfied once we can blame a human somewhere along the chain. But with ever more complex and deliberately unpredictable artificial intelligences, the day will come when we can't blame the human anymore. It is better to think now about whether it makes sense to blame an AI.

Verena,

your question is very interesting. I think a robot should have an ethics. See

http://www.intelligent-systems.com.ar/intsyst/ethartsi.htm

But, a person is responsible for the actions of his dog. Also, a person gives orders to his robot and will have to be responsible for its actions.

Further there is another question. What effect will intelligent robots have on human beings and human society?

My answer is that it will raise the level of wellbeing, see:

http://www.intelligent-systems.com.ar/intsyst/effects.htm

We always build better machines, without a limit. I believe that we will build more intelligent robots, because they will be more useful.

If cognitive automation can be developed to the point that it can evolve independently I can see it evolving beyond human cognition because each generation can simply download the knowledge and experience of the previous generation, given generational compatability. Each generation isn't required to re-learn everything from the ground up. In that case it may be necessary to treat an artificial intelligence as a sentient creature with the right to evolve without limitations.

Large databases aren't self-evolving. I don't believe cognitive automation in any form deserves recognition as a sentient entity unless it has the capacity to develop independently without interference.

So will the database be able to provide for it's own care? Can it perform self-maintenance? How do you define "self-evolve?"

Ah, a question I myself have been wondering about for a long time. With the work on Atomic Computers being done, AI systems are becoming a real possibility. Any harm that comes to a human being from a robot at this point in time is always opperator error. Telling the robot to go Home rather than to point X could result in a messy death. But it would never be considered the robot's fault. But what if we broke the self-conciousness barrier? Into the realms of existing. The moment a robot can look at itself (Not being ordered to by an opperator's programming) and think "I exist," is when fault comes into play. So, the real question is this: If and when robots become self aware, should we impliment something like Asmov's Laws, or should we simply let the robots evolve and live like we would fellow Carbon based life forms?

I myself believe that if we were to impose Asmov's Laws onto a robot, that the robots would eventually overcome those laws anyway for the very reasons Scott put forward. And because we had put forward those laws, the robots would mistrust us right off the bat. It is a sticky situation. I vote for simply letting them go, and integrating.

As for pulling the plug on an AI robot, one should look at it the same way one looks at the death penalty, or putting a pet to sleep. A truly intelligent being deserves the same justice as any other, despite it's composition. Chrome or skin. LED's or hair. Photo-receptors or eyes. It should all be counted the same.

Of course, for the moment this is all hypothetical. But it does no harm and may do quite a bit of good to think about.

Robots or Androids with consciousness have been the grist of many sci-fi stories with Azimov giving the subject more thought than most authors. I have also struggled with the Android question in several of my science Fiction stories (under my AKA Kalifer Deil).

We will be confronting this question in the face very soon. So far the military has not given it's robot planes the trigger finger but as soon as the planes have better target recognition than people, then I think it's a given, that it will happen.

The robot soldier is also not far away. An improved power source (10x better batteries?) are probably required. They will also happen and they will be far more interesting since it's likely that some will look like people to mingle with the enemy. In the above cases no consciousness is necessary, just locomotion, navigation and target recognition.

The real test is conscious robotics. This will be a test of the technology and a test for us human beings. We are beginning to understand the functions of the various parts of the human brain and what constitutes consciousness. This research is likely to follow several tracts. One tract is military robots using only what is necessary of consciousness to accomplish its goal. A second tract is building intelligent systems using what we know about the brain to capture that in a computer. This will use computer algorithms the replicate the various part of the brain without actually simulating those parts at the neuron level. A third tract is a detailed simulation of the human brain using the connectome of an actual human. Suoercomputers are a hairs-breath away from the speed necessary to do that in real time. Getting the full connectome of the human brain will take a little longer.

The second tract will be compact enough to put into an android and is probably less than twenty years away but the third tract will likely take another 15 years after that to become compact enough to fit in a human sized body. The third tract is the most interesting. People that will be able to afford a technological afterlife may not be the nicest people you would want to have around. You can't kill them because they could keep a backup. You can't incarcerate them because they could just turn that instantiation off and create another. In very little time they would become far more intelligent that humans and could exploit this to their advantage. Us human will quickly pass laws out of fear that restrict androids, limit their participation in the affairs of man, mark them for easy identity and kill them without legal recourse. As more androids are produced the tension between androids and humans will increase until a war takes place.

This can be avoided if great care is taken in android creation. Human's that become androids must be generous, loving human beings to begin with. It will only take one bad apple to swing public opinion against them since there will be religious prejudice against them from the start. Right wing preachers will start it off by calling them the devil's minions or golemm .Man creating an entity in his own image is a golem in the Judaism, Christian and Islam religious traditions meaning an android is going to have a tough time getting respect even if it is brilliant, is angelic, saves the lives of a thousand people and is full of love. More that half of the people alive today will see this happen in their lifetime. It will be an interesting time.