The two-edged knife of artificial intelligence / technophobia and technophilia are not wise.
According to Khabaronline News Agency, Reza Dastjerdi wrote in the introduction to his interview with Alireza Seqatoleslami for Ibna’s Religion and Thought service: “Ethics of Artificial Intelligence: Principles, Challenges, and Opportunities” by Luciano Floridi, a prominent Italian-British philosopher, founder of “Philosophy of Information,” and one of the most influential interpreters of the digital revolution, was published in 2023 with a translation by Alireza Seqatoleslami, a faculty member at the Iranian Research Institute for Information Science and Technology (IranDoc), as a recent publication by Kargadan Publishing House.
The winner of the Grand Cross Knight’s Insignia in 2022, also known for works such as “Ethics in Information and Artificial Intelligence,” “Information Logic,” and “The Essence of Information Philosophy,” examines the philosophical foundations of AI ethics with a coherent argumentative approach in this book. What follows is the outcome of Ibna’s conversation with Alireza Seqatoleslami, who also has publications including “Translation of Selected Articles on Information Technology Ethics” and “Specialized Glossary of Information Ethics and Computing.”
Alireza Seqatoleslami
In your opinion, why has attention to ethics in the field of artificial intelligence become important today?
In response to this question, I must first note that the importance of discussing ethical issues arising from the development of artificial intelligence systems is not limited to the last decade, especially recent years. The history of these ethical issues and challenges dates back to the second half of the 20th century, which were somewhat scattered around the emergence of information and computational systems, often referred to as “computer ethics.”
In the late 20th and early 21st centuries, these ethical issues expanded and gained greater precision around the development of information and communication technologies (ICT) and their widespread penetration into daily human life. They were referred to by names like cyber ethics, ICT ethics, and generally information ethics. Today, these ethical challenges, with added importance, are focusing more intensely on the development of AI systems and their growing penetration into various aspects of life.
If I want to talk about the importance of ethical issues surrounding AI development within the framework of Floridi’s reflections, I must focus on the issue of “AI agency capacity” and the development of related autonomous systems.
This means that human agents with their ascribed autonomy, after four scientific revolutions (influenced by Copernicus, Darwin, Freud, and finally Turing), have encountered artificial agents (intelligence!) that possess significant autonomy (or, in a loose sense, self-governance). This technological transformation marks the beginning of many ethical concerns for human agents, who might one day consider these autonomous artificial agents their rivals.
Why can artificial intelligence be harmful to daily life, even while striving to improve it? And is it fundamentally possible to align artificial intelligence and ethics?
Whether the use of AI systems serves to improve life and social good in human societies or, conversely, leads to harm and social evil, is essentially the classic “dual-use problem.”
This issue is not exclusive to artificial intelligence; it applies to many simple and complex technological artifacts. A “knife” (as a simple technological artifact) can be used for surgical procedures that lead to human health and pain reduction, or it can be misused for murder.
The achievements of “biotechnology” (as a complex technology) can also be used for human well-being and generally for nature, or misused for biological harm. So, we can say that generally, the use of technological artifacts falls into the realm of ethical and unethical use by humans.
I do not intend here to deny the value-laden nature of technological artifacts and their intermediary role in humans’ benevolent or malicious use of these artifacts. Rather, the point is that “responsible design, development, and deployment” of technological artifacts and products, from the initial ideation stage to their utilization, must continuously adhere to the red line of serving humanity and nature.
This process applies fully to the invention, development, and deployment of artificial intelligence systems as well. For example, the “UNESCO Recommendation on the Ethics of Artificial Intelligence (2021)” is a valuable example for the responsible design, policy-making, development, and deployment of AI systems, providing significant ethical strategies for all stakeholders involved in the design and expansion of AI systems.
Today, numerous documents and research programs titled “AI for Social Good” (AI4SG) are being prepared and compiled at national and international levels, aiming to apply and advance AI to address social issues and improve global welfare and well-being. Ensuring their implementation relies on the professional and ethical responsibility of designers, managers, engineers, developers, and ultimately users of AI systems.
When we talk about “artificial intelligence ethics,” what exactly are we talking about?
In short, “AI ethics” is a subfield of applied ethics and technology ethics that addresses ethical issues and challenges during the design, development, and deployment stages of AI systems. However, if we want to define “AI ethics” in more detail, it is appropriate to examine this field of study at three levels:
- Cognitive level (pertaining to understanding AI),
- Evaluative level (pertaining to evaluating AI),
- Prescriptive level (pertaining to prescribing ethical guidelines).
At the cognitive level, we introduce the nature of artificial intelligence and why it works as it does. At this level, philosophical reflections on AI are very helpful and provide a suitable explanatory and analytical basis for understanding it.
At the evaluative level, we identify ethical issues arising from the development and deployment of AI in human societies (such as privacy violations; smart ubiquitous surveillance; bias in decision-making systems; educational and research misconduct in the use of large language models (LLMs); and other issues) and collect appropriate ethical principles (non-maleficence; benevolence; justice; autonomy; explainability) to examine the mentioned ethical issues.
At the prescriptive level, within the framework of the collected ethical principles, we prescribe necessary ethical measures and guidelines for addressing AI ethical issues and subjects. What makes Floridi’s book “Ethics of Artificial Intelligence” distinct from many other works with the same title in this field, I believe, are his consistent philosophical and analytical reflections at the cognitive level of understanding AI, which provides the necessary conceptual foundation for evaluating AI. Although he does not delve into the prescriptive level of ethical guidelines in this work, through multiple references to measures and guidelines in international AI ethics documents, he makes them more understandable and justifiable for readers.
Which aspect of modern life should be most concerned about the use of artificial intelligence?
Truthfully, answering this question is very difficult. As I mentioned earlier, today, artificial intelligence and the use of its technological artifacts have influenced and penetrated all individual and social aspects of developed human societies.
That in our country, certain areas like research and education influenced by the development of LLMs, or smart cameras and their ubiquitous surveillance, and sometimes the violation of citizens’ privacy, receive more attention, is not due to greater concern about the intensity of penetration of these technological tools in these areas, but rather due to our failure and underdevelopment in many aspects of life to utilize them.
In many developed countries where the opportunity to expand and utilize these technologies in areas like smart transportation, smart health, smart policing, smart education, and many similar cases has been realized or is evolving, social and ethical concerns exist in all these areas.
And precisely for this reason, these countries have become more aware of the importance and necessity of addressing the field of AI ethics and have well understood that delay and inaction in adopting appropriate policies and measures to realize AI in the service of social good will have sometimes irreparable consequences for both human societies and the environment.
Should we fear artificial intelligence, or do we need to learn to live with it?
In general, human experience, especially since the industrial revolution and the rapid growth and increasing development of technologies and their artifacts, has shown that falling into the dichotomy of technophobia (fear of technology) and technophilia (worship of technology) is not a very wise action.
Undoubtedly, today, imagining travel from one city or country to another in a limited time, without technological transportation systems like aviation or railway systems, would probably be impossible, or imagining human interactions and long-distance communications without ICT systems has become almost impossible.
Artificial intelligence and its capabilities are no exception to this rule after more development and deployment in human societies. So, in Floridi’s thoughtful expression, just as we learned over time that besides human agents in nature and the biosphere, other living creatures and agents like animals and plants also possess ethical status and dignity, and the world without them endangers our environment, today we must also learn that in the infosphere (of which the biosphere has become a subset), besides human and living agents, artificial agents also possess a specific and minimal ethical status and dignity as informational creatures.
Although they do not possess human ethical status and associated rights, due to their limited agency and autonomy (or loosely self-governance) and ability to make decisions and influence independently and act in the infosphere, they do not have zero ethical status. It is with the realization of such an outlook that human agents, alongside coexistence with non-human agents, can responsibly undertake the design, development, and deployment of artificial agents and enhance and ethically elevate our hybrid digital living environment (the link between environment and digital environment).
Luciano Floridi
Is entrusting important life decisions to artificial intelligence the right thing to do?
Let me begin the answer to this question with an explanation of an emerging concept. Today, both technically and ethically, a project called “Explainable Artificial Intelligence” (XAI) is discussed.
In fact, explainable artificial intelligence is a set of methods, models, and frameworks that aims to make the decisions and outputs of AI systems (especially those that are more important and sensitive in human lives) understandable, reconstructible, and questionable for humans.
Understanding how an AI system led to a specific decision or output has many benefits, both technically and ethically. Therefore, AI explainability can help developers ensure the intended decision-making and output.
Today, explainability of intelligent systems is of great importance in the field of AI ethics, so much so that many regional and international AI ethics documents, such as the “High-Level Expert Group on AI Ethics Guidelines” (HLEGAI, 2019) and the “UNESCO Recommendation on the Ethics of Artificial Intelligence” (UNESCO, 2021), refer to this issue as an ethical principle and define entrusting important life decisions as conditional on the realization of this principle.
In Floridi’s book “Ethics of Artificial Intelligence,” alongside the four ethical principles of benevolence, non-maleficence, autonomy, and justice, the ethical principle of “explainability” is introduced as a new enabling principle for artificial intelligence. Under this ethical principle, it is stated that AI in the service of social good has transparent processes, linking the capabilities and goals of these systems transparently; and the decisions of these systems are explainable as far as possible to individuals directly or indirectly affected by them.
Does AI ethics have meaning without serious law and supervision?
Undoubtedly, AI ethics lacks enforcement guarantees without serious law and supervision. However, this statement does not mean that AI ethics should be reduced to law and supervision alone. Floridi introduces three realms in his book for realizing benevolent AI serving society: Digital Governance, Digital Regulation, and Digital Ethics.
He does not consider the realm of digital governance to be solely focused on the government and sovereignty, but presents it as a multi-level and multi-actor governance (including governments, corporations, universities, civil society, and similar cases) that implements the methods of creating and enforcing policies, procedures, and standards for the proper development, use, and management of the infosphere (which includes human and artificial agents).
He does not consider digital regulation as merely a legal tool to contain this technology, but defines it as the binding translation of digital ethics (and here, AI ethics) into law and institutional supervision with the aim of flourishing and well-being of the infosphere. And finally, digital ethics is a normative framework that shows what is ethically good or bad in the infosphere and how to protect and enhance the quality, integrity, and flourishing of the infosphere.
Although these three approaches have different norms, they complement each other and should not be reduced to one another. They can be considered three overlapping circles, with two overlapping in some areas and all three in one shared area. The relationship between these three realms can be summarized as follows: Digital ethics determines what should be; digital regulation translates part of digital ethics into legal rules; and digital governance enables the realization and continuity of ethics and law within it.
Such a model also applies to the interaction between AI ethics, law, and governance. The footprint of this model can be traced in the “General Data Protection Regulation of the European Union” (GDPR), which is the embodiment of privacy and personal data ethics, as well as the “European Union AI Act,” which is the embodiment of AI ethics.
Therefore, it can be said that firstly, the realization of these three realms without accompanying each other is very difficult, if not impossible, and secondly, reducing the realm of governance to the official institution of power, reducing the realm of regulation to conventional and traditional legal institutions, and reducing the realm of ethics to inconsistent moral teachings lacking a theoretical view, will only lead to confusion and missed opportunities in realizing good AI for social benefit.
