Is There A Possibility For An

Ethical Artificial Intelligence?

Att. Nur Banu Kaylan Deligöz

  • What is artificial intelligence?

It is the branch of science that examines how artificial systems can perform higher performance levels in every cognitive activity than natural systems[1]. In shortly, artificial intelligence scientists are trying to produce intelligent machines that can work and react in a similar or even same way to the features inherent in humans. Especially in the last 5 years, the development of artificial intelligence continues rapidly; it is used in various fields such as voice recognition, detection of cancer cells and non-driver vehicles by means of tools / machines developed in this field[2].

 

  • The Effect of Artificial Intelligence on Humanity

It was not very difficult for the transformative feature of artificial intelligence to influence humanity and it did not take much time. The fact that the effect is a revolutionary, that every artificial intelligence product is an answer to a great problem of humanity has caused the artificial intelligence community, including producers and developers, to overlook critical points. The most important of these is that these machines, which are developed for the purpose of facilitating human life, contain and disseminate ethical rules that do not conform to the natural structure of humanity. It is a problem that humanity can solve with no doubt that whether these tools and machines created with human intelligence can cause harmful effects with intentional or caste-free methods for people.

 

 

  • What Type of Loss can Artificial Intelligence Cause?

According to the predictions made by the experts[3], artificial intelligence will lead to many shortcomings of natural life as it settles into the normal course of life. In the next 5 years, the most likely damage we will see is fraudulent activities facilitated by artificial intelligence. By means of fake audio and videos created by the products that can be defined as the black face of the technology, it is impossible for people to understand that they are not an unreliable smart device and maybe they are sharing some confidential information with these machines that they will not share with the closest ones. It will be possible to see the negative effects of artificial intelligence placed on smart robots and many other areas, for example, the exaltation of pornography by using the faces of the people followed by the society, unauthorizedly in pornographic films (deepfake technology)[4].

 

  • Can Artificial Intelligence Be Ethical?

A report entitled “Ethics Guidelines for Trustworthy AI” was published on April 8, 2019 by an independent group of people with expertise in artificial intelligence within the EU Commission. According to the report, a reliable product of artificial intelligence must be legal, ethical and robust[5]. The expectation that the law will grow in the speed of the development of technology is not a reasonable expectation as to the nature of the law and our past experiences so, it is a great need to put forward the ethical rules that artificial intelligence should comply with in all circumstances. In response to this need, the report sets out the principles of these three characteristics;

  • To respect human choices,

  • Avoiding damage / minimizing damage,

  • Be honest/ fair,

  • Be clear / transparent.

The report prepared in a graded structure regulates the 7 things that artificial intelligence should provide to comply with these rules and to have three basic characteristics that are initially included. These subjects will be briefly described in the next section of the article.

 

1. Human Agency and Oversight: The first feature expected to be carried by artificial intelligence products that are human creation; they support human autonomy, protect human rights and contribute to the creation / maintenance of a just society order. Artificial intelligence, which facilitates the use of many fundamental rights, such as the right to education, the right to vote and to be elected, may also have adverse effects on fundamental human rights. In the event of a negative effect, because it is very easy for smart devices to connect the Internet and to connect with each other, the spread of the impact will be very fast. Therefore, a fundamental rights impact assessment prior to each production / development work will be unavoidable to achieve this requirement.

 

2. Technical Robustness and Safety: A reliable artificial intelligence product must have algorithms that can provide a safe and robust return / remediation to its users in the event of any error and / or inconsistency during its use. This feature can be interpreted with the ability of artificial intelligence to reduce the risk of damage as much as possible and to minimize non-intentional / unexpected losses. Resistance to technical attacks such as hacking or data leakage, algorithms which are designed to correct errors, and giving the same results in similar uses (consistency) are indications that an artificial intelligence is robust and safe.

 

3. Privacy and Data Governance: EU General Data Protection Regulation (GDPR), and the Law on the Protection of Personal Data No. 6698 entered into force on 7 April 2016 in parallel with this regulation, enforces the obligation that all data processing activities that data managers perform with automatic or non-automated systems must comply with these regulations. However, the fact that artificial intelligence has not been fully established on legal grounds also limits the application of regulations in this sense. For the time being, this is the only way to ensure that the smart devices, which are easily connected to each other's systems and which are able to obtain personal data without the consent of the relevant person, are again produced to be respectful to the privacy of private life which is a fundamental human right. In this sense, data controllers should not only construct their systems in a way that cannot process unlawful data, but also take necessary technical and administrative measures against possible data leaks.

 

4. Transparency: Artificial intelligence products developed by people to make people use and ease their lives should have the designs that people can easily keep under control. In order to ensure transparency, all operations, primarily data collection and data identification processes, must be established through traceable algorithms. Humanity, which gives the ability to decide on a machine, should be able to understand the decisions it takes when the machine is operational and to manage the processes related to it.

 

5. Diversity, Non-Discrimination and Fairness: It is expected that human talents should be taken into consideration in the creation processes of artificial intelligence, and that accessible and usable systems will be created for all. In order to achieve this feature, smart devices have universal designs; without examining the age, gender, social environment criteria, they should be used by everyone. This can be considered as a precondition for the expected fairness of artificial intelligence.

 

6. Societal and Environmental Wellbeing: Intelligent devices that have settled in life can easily affect and even manage the changes in the social dimension. Artificial intelligence, which has such a great influence on social welfare; need to support positive transformations, environmental responsibilities and sustainability. Implementation of all stages of artificial intelligence from production to sale with care to save energy, the use of speed and advanced structures of artificial intelligence in the conduct of election processes in accordance with democracy and fairness are some of the examples that can be applied to achieve this feature.

 

7. Accountability: Artificial intelligence should be able to guarantee its responsibilities and obligations in all stages of production, development and use. Determining the results of the movements and decisions of artificial intelligence, analysing, documenting and finally minimizing the damages will reveal the positive and negative results that may arise from these systems and will be able to protect the individuals who are harmed by indirect ways.

As a result, the ethics of artificial intelligence will be possible only if these 7 rules, summarized briefly above, are evaluated and applied in specific cases. Isaac Asimov, the famous sci-fi author, says in his first law of the Three Robot Laws that; “Robots cannot harm people or become inactive and not allow them to be harmed”. This principle of explains the goal that these 7 items are expected to achieve. Artificial intelligence should be used and developed as long as it complies with the ethical rules that humanity has created for centuries and able to provide superior technological developments for the benefit of humanity.

 

BIBLIOGRAPHY

  1. Prof. Dr. Cem Say, 50 Soruda Yapay Zeka, İstanbul, 7 Renk Basım Yayım ve Filmcilik, October 2018

  2. https://beebom.com/examples-of-artificial-intelligence/

  3. https://www.dw.com/tr/yapay-zekan%C4%B1n-insanl%C4%B1k-%C3%BCzerinde-etkisi-ne-olacak/a-41522940

  4. Deepfakes Explained: The AI That’s Making Fake Videos Too Convincing, https://www.makeuseof.com/tag/what-are-deepfakes-explained/

  5. Ethics Guidelines for Trustworth AI, Independent High-Level Expert Group on Artificial Intelligence, European Commission, April 2019

 

[1] Prof. Dr. Cem Say, 50 Soruda Yapay Zeka, İstanbul, 7 Renk Basım Yayım ve Filmcilik, October 2018

[2] https://beebom.com/examples-of-artificial-intelligence/

 

[3] https://www.dw.com/tr/yapay-zekan%C4%B1n-insanl%C4%B1k-%C3%BCzerinde-etkisi-ne-olacak/a-41522940

[4] Deepfakes Explained: The AI That’s Making Fake Videos Too Convincing, https://www.makeuseof.com/tag/what-are-deepfakes-explained/

[5] Ethics Guidelines for Trustworth AI, Independent High-Level Expert Group on Artificial Intelligence, European Commission, April 2019