it depends, fuzzy logic can help us, knowledge-based systems can help, but moral questions can be answered in different ways, from different perspectives, however, it should consider all aspects and details. I think it is possible now to design a system for answering moral questions, but it cannot (and will not) satisfy all humans.

Hi Ali,

No human can satisfy all humans. Nature of humans psyhology. I have theory: in moment when Ai can generate/create joke that is funny to humans, then that AI is ready to help us with some moral questions

Dear Dr. Mlakic, Would you please give more details on what you preceisely mean by saying: "No human can satisfy all humans." Furthermore, without putting into a relevant context the phrase "generate/create joke that is funny to humans" is not obvious. What is joke for one, might not be the same for other. Anyway, jokes and moral questions are very different things in nature and role. Best Regards, I.H. 

"No human can sutisfy all humans" means whatever you do there will be someone sho is not confortable what you do. There have to be some line drawn for AI what to do and what not. That line we humans have to find but question is can technology help us in that?

"generate/create joke that is funny to humans" is generaly speaking AI generated funny story to humans who can hear or read that joke in that moment. Not going in discussion about jokes just capability of AI to generate jokes

Dr. Mlakic,

You are bringing up a very interesting discussion topic: delegating human (behavioral, ethical, social, etc.) norms to intelligent artifacts and systems. The issue is: most of the norms are de facto norms rather than de jure norms, with a strong cultural flavor and context dependence. Therefore, I do agree with you, this topic needs further insightful discussions. But obviously, I do not have any strong idea about a solution considering this matter that is also philosophically interwoven.

Kind regards,

I.H.

Dr. Mlakic,

Linguistics, anthropology, computation, psychology, physiology, etc. have intensively studied the phenomena of joke and joking (as manifestation of humor). Current research is however more about interpretation, classification, and measurement of humor than about transferring humor to artificial things and dealing with reflexivity. Are you suggesting that instead of, e.g. proposing the 66th psychological measurement instrument, we have to establish a new discipline? Mind, I am not joking with this!

Best regards,

I.H.

Agree with Imre Horvath. Moral questions can be answered by AI, but its answers cannot be better than answers provided by experts in some complex cases like crime scene cases. Answering moral question is possible now, in other cases, there are also many robots that can tell jokes (or chat with you http://chatwithigod.com/) but your idea cannot  be a new revolutionary thought for future.

Also consider this fact that some questions have no answers: for example: can a terrorist do a good job for his society indirectly? The answer is "no" to what degree? Consider that person is a scientist in biology and makes bad guns, is he a bad guy? Some persons have no choice, because of poverty, and they should work as drug dealers, but still they help other poor people by that money, is he a bad guy? Is helping by that money is a valuable work? AI cannot find any answer for this kind of problems, because humans cannot (should not be certain about their information) judge and then easily (by narrow-mind) put a positive value for some works or other humans. Now, AI should do that and what should AI say to us? it is complicated. Best wishes.

It will be very difficult, if possible at all, for AI or whatever to get to the very depth of humanity,

Answers provided by AI about morality cant and will never (i suppose) be better than asnwers from human experts, just because those questions are relative. They change from time to time and from one place to other. Just considering those two dimensions we have infinite (almost) pool of answers. This problematic will shurely be on going discipline. There is another question rising up from this, and I think its much more important for us humans: "Can AI research give us more knowledge about us (humans) than we know now?"

Dear Mlakic,

              Answers to moral questions are highly subjective. Is it moral to feel human beings are superior to other animal species? What right we have to consume resources to wipe out many species? Do we have right to poultry farming by raising chicken in cages? We choose to ignore inconvenient Truths to many moral questions.

             However, I don’t believe real Artificial Intelligence can be invented in our life time. Software engineering knowledge is so primitive. Today software researchers are clueless  about the nature, properties and functioning of good old components to invent real-software-components for achieving real-CBD for software. If they can’t even understand such simple things, how could they understand the nature and functioning of neurons and neural networks (essential for real-AI)?

              For example, a small fly of size 1-cubic millimeter can navigate (e.g. avoid obstacles) better than the best self-driving cars in the world. Even supercomputers can’t fly a jet-fighter better than the fly (having few thousands neurons).

Best Regards,

Raju

Dear Dragan, a very good point is in your last reaction. Perhaps an AI with a very different or largely extended focus and aim ... (but we should most probably call it meta-AI or something like this). As Raju argued, there is both an undeterminism and a stong limitation out there, beyond the human context  Regards, Imre

Hi Raju,

I think we will see some kind of AI authonomy in oure life time, but I cant say in what form. Development of general AI is ongoing research never stoping and past few years its in focus of some Universityes and research centers (MIRI e.g.) and they recruting new and fresh minds all around world. Thats impressive to me. Even if we think that some actions and skills are reserved for human beings, we can face new and unknown field of Artificial Creativity which is ongoing research. 

Dear Imre, 

Thank you, I think am on right direction here.

i think limitations just make us feel safe and secure in oure shell, but that shell is long ago beyod oure capabulities. Limitations are relative. Just look at past 5 years in HW and SW. Its huge jump.

Dear Mr. Mlakic,

            Based on my experience with the research community, I am not very optimistic that real AI could be possible in our life time to the level that can debate complex subjective subjects such as morality. Today AI is nothing more than expert systems driven by algorithms. We have no idea how neurons work in neural networks.

            I have been struggling for many years to convince research community to investigate facts, objective reality and evidence to discover nature such as true essence of CBD for physical products and essential properties uniquely and universally shared by each and every known physical component. Once this kind of BoK (Body of Knowledge) is acquired, it is a trivial task to invent real-software-components for achieving real-CBD for software, which addresses the infamous software crisis by eliminating spaghetti code.

           Today software engineering is in crisis because software researchers defined nature and properties of components and CBSD without any basis in reality. They have been doing the same mistake in the case of neurons and neural networks. Is it possible to invent cures for infections, if scientists blindly define nature and properties of versus and bacteria (without giving any consideration to reality or facts)? If it is wrong to blindly define nature and properties of versus or bacteria, why is it not wrong to blindly define nature and properties of components or CBD (without any basis in reality/fact)?

            Is it hard to understand this simple logic: How is it possible to invent fiber optic networks, if the researchers are clueless about the properties of light and how it behaves in fiber optic strands? How is it possible to computer chips, if the researchers are clueless about the properties of electrons and how electrons behave in semi-conductor material?

             How can researchers invent real-software-components for real CBD if researchers blindly define nature and properties of components and CBD, not only without any basis in reality/fact but also in clear contradiction to the known reality and facts about the physical components and CBD of physical products?

            Please look at the number of neurons in fruit fly: https://en.wikipedia.org/wiki/List_of_animals_by_number_of_neurons. Today we can build 10 billion transistors in a CPU. I am sure, we have more than enough capacity to emulate the brain of a fruit fly, by employing firmware and software.

            Of course, we need to discover how neurons function to emulate them. Although it is a trivial task to make GUI-platforms that can create real-software-components, today no GUI-platform is capable of creating real-software-components essential for real-CBSD. Likewise, it may not be hard to emulate neurons and neural networks, once we acquire the BoK about the functioning neurons and neural networks (for simple things like self-driving cars by emulating the brain of fruit fly). Of course, it would be a long way (many more decades) from able to engage in philosophical debate about the morality.

       Unfortunately, after struggling for many years, I am not even optimistic that I can convince software researchers to gain BoK about the components and CBD with in my lifetime. I created comprehensive proof, technologies (e.g. obtained 6 US patents so far), but no one even willing to look at the evidence.

Best Regards,

Raju

Hi raju,

Am glad that one of first researchers in CS is answering me. I am honored. Thats exacly point what you are saying! many of us that are not involved in CS (Computer sciece) need to be involved in AI research, not just because manpower and "interesting robots", but also because depth of knowledge that needs to be involved for succesfull and faster research in that field. Same problems are present in all branches of science, because we are involving people with academic knowledge from on narrow field who doesnt understand nature of things that are involved in concrete research. Thats one of reasons, maybe major reason, why I think that science (AI ) may give us clue about some other questions that are not particulary technical nature. We scientists dont have luxury to compete, we need to put job/result in focus and commit oure resources to that goal, that includes other branches of science and other people willing to get involved.

Thank you again

Dear Mr. Mlakic,

            Thank you for your kind words of support. This kind of moral support and encouragement gives lot of strength and motivation to continue my struggle to compel research community to investigate evidence, reality and facts to discover no hard to achieve real-CBD (Component Based Design) for software, which can be summarized as: CBD is Implementing about 90% of the features and functionality in replaceable components, where replaceable component imply components that can be easily dis-assembled (or un-plugged to redesign or refine individually outside of the product free from spaghetti design/code) at any time for periodical feature or functionality maintenance or upgradation releases and re-assembled (or re-assembled after testing it individually free from spaghetti code). Since over 90% of the design and code for features and functionality is implemented in such components (where each is free from spaghetti code), over 90% of the design and code is free from the spaghetti code.

Such real-CBSD addresses the infamous software crisis, because the root cause for the software crisis is the infamous spaghetti code. Design and development of no other product is infected with such spaghetti code, for example, particularly design and development of complex one-of-a-kind-products such as experimental spacecraft or fully-tested pre-product golden prototypes of next generation Jet-fighters. It is not hard to achieve such real-CBD for software by using real-software-components:  https://www.researchgate.net/publication/292378253_Brief_Introduction_to_COP_Component_Oriented_Programming

Today no one even knows what the true essence of the CBD is, but every researcher insist that it is impossible to invent real-CBD. I have been struggling to provide counter evidence for many years such as hundreds of software applications built by assembling replaceable components (e.g. See real-software-components.com).

My feeling is that, real-AI can be achieved by emulating neurons and neural networks in the brains of smallest insects to perform specific tasks such as driverless cars. To do that, we need to invest substantial amount on the research to understand the functioning of each neuron and they collaborate and function in networked neural networks.

The clock speed of brain waves is not very high (i.e. about 100Hz), while CPUs in PCs run at about 3Ghz to 5 Ghz.  This suggests, we can create a multi-core CPU (about 250,000 cores), where each core (having few thousand circuits with embedded firmware) to emulate a neuron.

It is a trivial task to invent real-software-components for achieving real-CBD, once basic BoK is discovered about the nature and properties of components and CBD are discovered. Likewise, I feel, it is not hard to build neural networks, if we can discover the functioning of each neuron and how they collaborate in the neural networks.

Of course, we need to discover how natural instincts are programmed into the neural networks and how neural networks are formed based on each of the experiences. I feel, who we are, is determined by our natural instincts (inherited from the birth) and are also shaped by every day our experiences. I remember reading somewhere that, many species perform complex tasks naturally without any training (e.g. weaving a nest by sparrow).

Each ejaculation delivers 100 million sperm (in about 1 ml). Each sperm has code for creating a complex human being. Doesn’t it show how insignificant and primitive our software code is? It is humbling, isn't it?

Best Regards,

Raju

Article Brief Introduction to COP (Component Oriented Programming)

I will try to keep it short and simple , the answer is no for the artificial agents that have been developed till now , because we simply do not understand how to mathematically formulate or describe moral values and emotions , but once if some day an actual AI system gets developed it may be able to add emotions to itself by figuring itself out that what emotions are and what moral values are and how to formulate , express and code them!!

Dear Dr. Sharma,

Shall the capability of meta-artificial intelligence (M-AI) systems to add emotions to themselves by figuring out what emotions and morality are, what emotions and morality apply in various situations, and behave accordingly be important for the M-AI systems in the same manner (with the same teleology) as it is for human beings and communities, or only for humans, or for both? I am interested in your opinion, also with regards to what this capability means, how could it be achieved, and what impacts will it have on human emotions and morality? Can M-AI systems be believer or religious?

Kind regards,

Imre Horvath 

Lets not go into morality of humans that's much more than we need here. AI will have to adapt, or we will, to society where humans are dominant. Questions that involve human behavior are questions that AI need to answer in real time. According to its training, AI will act as intelligent being in specific place on specific time. Can that behaviour help us humans understand ourselves little bit more? Can AI in its implementation help us know US better? AI will give us straight answers and they are based on pure logic (or not is training was badly deducted). Situations where we will find us questioning are something like this: what will we ask AI who has Facebook profile for some time and has learning humans through it?