Let me tell you that not only machine intelligence, but machine implementation in general conceals pros and cons. An airplane with passengers is a machine that may not exhibit AI and nevertheless, may behave not as intended and malfunction causing loss of life and injuries. But the machine has no values as "values" is a category not applicable to machines, but to humans.

People decide whether to implement a given technology or not to implement it. No airplanes -- no airplane crushes. It is obviously a matter of choice. But on a global basis of the human civilization one thing seems to be inevitable and out of choice -- the result of the social natural selection that according to many humanists such as Kurt Vonnegut will eventually lead to dehumanization of the human population. The social natural selection is the result of the human choice on a global basis -- democracy. Another aspect of the same inevitable problem is the major driving force of the Earth life -- greed.

This is not a comprehensive answer, I just wanted to clear some misconception about ML and pave the way for a larger debate. I will partially repeat my own answer to a question concerning a similar issue, asked by Tony Maine: https://www.researchgate.net/post/How_to_discover_what_a_neural_net_knows

Machine learning algorithms do not develop any independent internal objective. All of their behaviors are defined only by the external objective imposed on them during training by the humans who design them. Therefore it is a fallacy to blame a ML system itself for making some decision, just as it is a fallacy to blame a car for running over a pedestrian. You can either blame the driver whose mistake led to the accident, or the designer who failed to spot a construction flaw that allowed that accident to happen. An ML system is alike in this aspect: if it "makes a decision", it is only because it was taught to make such decisions. This behavior was rewarded during training as it reduced the cost function which those algorithms seek to minimize. Ultimately however, it was a human who designed that function and selected the parameters that eventually led to the system learning to behave in a specific way. That human might not have been able to perfectly predict what exactly will the model learn, the model itself might not have been accurately validated and thus it may fail in some cases, but these are merely technical obstacles that will be eventually solved in the future. In the end, what matters is that there is always a human responsible for the actions of a system they design or use - no matter how complex the system.

We as a technical society are currently wielding more and more powerful tools. Some even allow us to measure our own "value" and assign it a number - and as your example shows, decide who is worthy of a treatment and who should be abandoned. But the responsibility for those decisions cannot and should not be hidden behind complex algorithms and computer systems. We as their designers and users will have to find answers to those deeply moral questions, instead of blindly following our tools and trying to blame them.

Hello,

This is a very interesting topic, which I was thinking about for the last few months. Basically, the application of AI on big data analysis was not intended to discriminate or classify people. However, in the world of business, some profit plans may lead to such act. But, it is the authorities reponsibility to request the codes used and analyze the results before allowing such applications to be put in use. AI and big data are not a reason to validate any act or rule.

Thank you

Thank you, Svetoslav, for your well thought out detailed answer. A gloomy, thoughtful view. And unfortunately, one can’t say that it is excluded that human negative behaviors prevail. But, are you sure that people will still decide whether to implement a given technology or not to implement it, if the AI is well developed in future? And, if people will decide not more, whom does that do for? best regards, Petra

Thank you, dear Przemysław, for your very interesting answer. And thank you for not responding comprehensively, because then the discussion would have been over :-). I fully agree with you concerning the relationships and responsibilities of human, machines and algorithms. And a fully agree with your opinion “We as their designers and users will have to find answers to those deeply moral questions…”But, are you sure that people will still decide whether to implement a given technology or not to implement it, if the AI is well developed in future? And, if people will decide not more, whom does that do for? Best regards Petra

Human beings will continue to improve Artificial Intelligence. The created will never become better than the creator!

Hello,

My first answer was for the near future. However, as a futuristic approach if humans succeeds in creating an autonomious, self and environmetal awear AI system, then many questions are essential to be asked by researches of that time. Furthermore, we will not know how the responce of that AI system will be since it doesn't have the basic instincts of biological creatures that affect their decisions.

Thanks,

Dear Petra,

Humans will always have the right to decide until... and if... they loose their right at all, i.e. if humans loose the power to overcome the machines they've created. Consider this simple example. A human creates a locking mechanism for their garage door, but, unfortunately, they get locked inside and the locking mechanism malfunctions. Then the game is over. It is not a question of AI, it is a question of machines, AI is just a kind of a machine, not a new realm. If we go further in our philosophical generalizations we could, on a well ground, state that machines are a realm belonging to an even more general set of rational objects -- the determined cause-effect material world itself. If not alerted, the human race may be "excluded" from existence through cause-effect sequence of events, in the beginning of which stays a human action -- another analogy: one shoots with a pistol into a titanium plate, the ricochet kills them, the pistol-plate system is a machine too :-)

Your question is extraordinary, and has become the major thematic line of one of the greatest films ever made. I mentioned this film in my previous answer -- 2001: A Space Odyssey. The machine is at loose, the human is "locked in the garage", then the humanity has one and only choice -- to become what it really is -- irrational, or die. Through irrational means Dave overcomes the machine, but at a cost. Humanity has overcome not only HAL9000 but the rational reality of the Earth world, which places the human Dave into the non-Earth true world and he leaves reality.

Remember John Steinbeck? " There's a hole in reality through which we can look if we wish".

:-)

Dear Svetoslav, your thoughts are interesting and profound and these thoughts touch me and inspire me to think further. However this does not lead to any optimistic conclusions…I looked at your RG-Page. You obviously very well about the topic. Can insiders, who are also human beings, help solve the problems in the interest of humanity? Of course I know that the total amount of humanity consists of different of very different individuals with different goals...

Dear Petra,

I am optimistic. I can't add any further to this topic :-) Maybe to other topics that you may raise :-)

Dear Petra,

Thank you for the question, it is really good that everyone feels sensibilized on this subject: and thanks for sharing a direct experience of yours so important.

AI bring the greatest change to which humanity is approaching in the very near future, and that change proceeds at great speed.

A first answer is representend by a question like yours, cause it's essential to be aware of the existence of the matter.

A second answer is in our educational system, which will have to take responsibility to increase the awareness, especially of future generations: not so much on technical aspects (in order to satisfy the market) but on ethical ones, in order not to lose sight of the primary objectives of society.

Thank you and see you soon!

Marco

Thank you, dear Marco for your answer and the good words. Yes, you see the topic is very important for me and really moves me. I hope that as many people as possible will think about the topic and discuss about it. Perhaps then negative effects and dangers can be identified and avoided. You are right, Education and mediation of ethical values can be an important building block. Petra

Petra Seidel

you're welcome :)

Dear Petra,

The learning machine (ML) is an advanced statistical calculating machine, which works on a collection of experiences that are inputted into a computer system.

ML works using the principle of similarity from two experiences

ML will be able to find better similarity, if the amount of experience included also increases.

So if your mother's case experience handled by your brother's doctor is put into the ML system, then in the future, the ML, (as a tool) will provide a more suitable answer.

Thank you, dear Iping, for your helpful explanation. This will be definitely be the case in this particular case. Assuming somebody enters the changed data into the system. But are you sure, it will still be the case, when most of the processes in society happen in the future through artificial intelligence and machine learning? For example, it is usually no longer possible to validate prediction or forecasting models for the occurrence and development of pathogens and pests of plants, used in German agriculture. These models are the basis for using of chemical and biological plant protection products. Without consistent validation and evaluation such processes can now longer work properly. How will control will be possible in the future, when most of the processes in society happen through artificial intelligence and machine learning. Who controls whom?

Best regards, Petra

Hi @ Petra Seidel. I think that the problem of seeing the treatment to cure a person only from the point of view of utility is cruel, but it's not correlated or dependent on what has been called Artificial Intelligence.

Thank you for your answer, @Roberto - Mineado. But such mathematical methods, procedures and algorithms form the basis for the development of artificial intelligence. The purpose of the discussion is not to avoid AI. The goal is to draw attention to possible problems and undesirable developments. Anyway, as long as man can intervene in this process, as long as man can still control him. The story of my mother’s stroke was just one example to illustrate negative consequences. This can happen if ethical principles are ignored. By the way, nobody told us in 2011 that the decision was made by means of machine learning. I did not hear that until September 18. In 2018, in a lecture in which a mathematician explained for 2 hours, what machine learning is, how it works, and that was one of the examples he talked about because it was used in some university clinics in 2011. Do we really know in which areas of life decisions are already made by means of these methods? Do you know the book: Cathy O’Neil Weapons of math Destruction: How Big data increases inequality and threatens democracy” mentioned above? Best regards Petra

Many thanks to all who participated in the discussion with their inspiring answers. You have fueled the discussion. And thank you all silent readers for their great interest in the discussion. Best regards Petra

As ML/AI used and development increases, I believe that it is best to include an aspect of human in the loop to help with decision making. In certain case (as the example in the question) a tailored assessment is needed (in my option) to complement the results from ML/AI, and this will then be fit back to the ML/AI module to make the systems better.

Thanks Abdullahi Arabo Abdullahi Arabo for that interesting reply that sounds good and could help. Is that an idea of you or will it already done? So one can implement or apply it already? Best regards Petra

Petra Seidel This is something that I am currently working on, but mainly applying it to a different case study. If it is something that you are interested we can share ideas and possibly collaborate.

Human intelligence can never be replaced. Artificial intelligence aimed at performing routine but specialized work performed by humans.