Home » Conferences » About Machine Ethics and Artificial Intelligence Safety


May 2018
« Apr   Jul »

Read these articles


About Machine Ethics and Artificial Intelligence Safety


We are all familiar with the movies in which robots take control over humans and then chaos is brought into the world just because these forms of artificial intelligence became too developed. But is this actually possible or is it just an invention of the movie directors in order to attract more audience? Researcher Utku Kose tried to depict the possible scenarios in which this could happen and he also came up with some solutions in the article Are We Safe Enough in the Future of Artificial Intelligence? A Discussion on Machine Ethics and Artificial Intelligence Safety. Starting from the point that currently there are many anxious people about the possible existence of dangerous intelligent systems, Utku talked about many hypotheses and tried to find some answers to questions that probably we all have asked at a point in our life.


Can machines be ethical? Machine Ethics is a research topic in which researchers’ deal with ethical issues that may appear while intelligent systems are solving some problems. These problems are generally associated with the existence of humankind or need very critical analyses to provide the most accurate solution at the end. In the literature of Artificial Intelligence, a similar term: Ethical Artificial Intelligence can also be used to meet the same research efforts, but Machine Ethics is focused more on an exact intelligent system that is something like a robot or an advanced machine, which is strong enough to solve more complex problems that cannot be solved yet with current capabilities of intelligent systems. The questions that emerge are first related to rights. “What kind of rights should an intelligent machine have while solving problems for us? If an intelligent system causes an accident or show dangerous behaviors while working on its tasks – jobs, which rights should belong to it in order to have accurate judgments about it?”. Of course, then we can take into account duties. “Should machines have the same duties as humans?” Then the author took into account human welfare. Keeping human welfare at an optimum level may be a big problem for robots because it includes an almost infinite number of environmental factors to be evaluated because of dynamic nature of the human and also factors (i.e. nature, world), he / she is actively interacting. When it comes to justice, the questions that emerge are related to the ability of robots to apply justice in different situations. Can an artificial intelligent machine consider what is right and what is wrong and then apply it in a certain context? “What does it mean for a robot to be fair when a duty assignment problem should be solved ethically considering all other robotic workers’ past performances at an intelligent factory?”

Furthermore, researcher Utku Kose talks about the three concepts that are quite important to this matter: information, control and reasoning. The subject of artificial intelligence safety is also addressed, being associated with the idea of that it is possible to encounter with unsafe – dangerous intelligent systems if effective precautions are not taken. The idea of a red button that could turn off a machine (if it is not working properly anymore) is also taken into account. When we talk about the future of artificial intelligence, the author states that “today, there are different kinds of research interests introduced in the context of Future of Artificial Intelligence. With their remarkable and logical background shaped by both technical and social aspects, two of them attract researchers interest widely. They are the technological singularity and the superintelligence” (Kose 189). On the first hand, technological singularity is a hypothesis indicating that Artificial Intelligence, which is better than human intelligence will shape the humankind and civilizations greatly in the future. On the second hand, superintelligence is both a term and a research interest, which is explaining an intelligence, which is surpassing even the highest level human brain in the context of general intelligence.

Kose also talks about the existential risks and the possible scenarios causing questions. Some of the scenarios could refer to the moral dilemma (Ex: How should a person be punished by the intelligent judge, if that person has killed somebody in order to save himself / herself?) the artificial intelligence created by artificial intelligence, the jobs that could be performed just by the machines, the rights and copyright problems or the social life transformed by digital individuals (Ex: ‘John meets every day with his virtual friends to discuss some Philosophy and Rock Music. He does not need any human friends because his virtual friends make him improve his knowledge.’).

In conclusion, Utku Kose offers some possible solutions to the problems that may be encountered when talking about programming these artificial intelligent machines. Along the study, the dangerous scenarios are also tried to be met with some possible solution suggestions. The study has been also organized as a reference work for enabling and directing interested readers to deeper sides of the literature. s. It seems that improvements in supportive technologies will possibly cause also many rapid and advanced developments in the field of Artificial Intelligence, but still algorithmic and mathematical structures have important role on solving even the most complicated issues that we may encounter with even in the future.

Ecaterina Filimon