5. Ethical Principles, Benefits and Issues of AI

April 8, 2022 lectured and written by Dr. Merve Ayyüce KIZRAK
“It is change, continuing change, inevitable change, that is the dominant factor in society today. No sensible decision can be made any longer without taking into account not only the world as it is, but the world as it will be.” — Isaac Asimov
This week, we held our class face-to-face at Bahçeşehir University Beşiktaş Campus. It was great meeting 40 very bright students. We have recapped the past 4 weeks. After remembering the definitions, concepts, and use-cases, we played a game to remember all the information easily. Thereupon, we identified the benefits of ethical principles and the issues we faced, which is the topic of this week. Then, we discussed AI in the Court of Law, one of the Unesco AI Ethical Dilemmas, in the midterm.
In this lecture, we discussed and continue to discuss the ethical issues posed by the development, deployment, and use of AI. Let's start with a review of the ethical benefits of AI.

Ethical Benefits of AI

When we talk about AI and ethics, we often encounter bad moral examples. However, there are also benefits that AI promises. As mentioned in many policy and strategy documents, it is expected to have economic benefits due to its positive effect on productivity and efficiency. It is foreseen that society will lead a better life and increase their welfare. The EU's High-Level Expert Group on Artificial Intelligence makes this very clear when it states:
"AI is not an end in itself, but a promising tool to increase the development of more people, thereby improving individual and societal well-being and the common good, bringing progress and innovation."
When we consider the field of AI, we cannot ignore the technical capabilities it offers us. The International Center for Risk Management highlights the ability of AI to process and analyze great amounts of data quickly and simply. Its prominent advantages are that it learns the patterns in these data, connects them, and makes new discoveries. In addition, AI can be more consistent than humans and adapt to changing inputs. In this way, we can say that AI helps people in the sector in which they are technically applied.
Ethical benefits that emerge as an effect of AI's technical capabilities can also be listed. These studies, often presented under the heading "AI for Good", have spawned numerous attempts to identify shared ethical values or goods such as benevolence, safety, achievement, and self-management.
Two different approaches can be observed towards identifying the ethical values ​​that AI must promote in order to be considered AI for Good: substantive goods and procedures for achieving them. The UN's Sustainable Development Goals and 17 comprehensive sets of goals have been described as "the world's best plan for building a better world for people and our planet". SDGs are interesting from an AI ethics perspective because they can be understood as the closest thing to human consensus for moral purposes. That is why it has been accepted by the UN and most member states. Moreover, they are not only promising but also measurable with indicators. Therefore, one of the most visible initiatives by the UN's International Telecommunication Union to promote AI for Good, one of the titles of the AI ​​for Good Global Summit series is “Accelerating the UN SDG”.
There are also data and analysis tools that Facebook aka Meta makes available for disaster, epidemic, population, hunger, and similar social crises and so on. Since this is demographic data, we discussed it in class next week to ask the expert while talking about privacy preserving techniques. Their motto is
"We empower our partners with data that empowers communities, improves social issues, and protects privacy." — Meta
Let's try to understand the issues through case studies to determine which issues people perceive as AI ethical issues.

Ethical Issues of AI

AI and ethics is a topic that has been discussed and studied for years. According to the research and some inferences, it has been systematically shown how people who work and/or use AI perceive their AI and ethical concerns. The case studies have a methodology that seeks answers to the "how" and "why" questions of facts and events that the researcher has little or no control over. Among the broad application areas of AI, the following are within the scope of research:
  • employee monitoring and administration
  • government
  • agriculture
  • sustainable development
  • science
  • insurance
  • energy and utilities
  • communications, media and entertainment
  • retail and wholesale trade
  • manufacturing and natural resources
In one study, two institution members, at least one technical expert, and a participant with management/organizational expertise were interviewed for each case. The case study includes 42 people [1]. A questionnaire study with Delphi approach was also carried out on the same group. This group includes all stakeholders of the industry, including technical experts, industry representatives, policymakers and non-governmental organizations. The questionnaire was sent to 250 experts working in the field of AI and ethics. There are 41 questions in total in this survey. As an example of the questions: “What do you think are the three most important ethical or human rights issues raised by AI and/or big data?” The survey findings and the results of the case study were combined within the scope of this research [2]. Let's examine the results!
As a result of the studies, 39 different ethical problems were identified. When we classify them according to uses of the term AI, some can be grouped under Machine Learning (narrow AI) and some under the more advanced and general AI scope we have not reached yet. A significant amount of ethical issues can be characterized by living in the digital world and socio-technical systems in general. Essentially, this classification is the distribution of ethical issues to AI concepts, as seen in the figure below.
Concepts of AI and ethical questions they raise
Let's list the identified ethical issues under three concepts of AI [2].
As a result of case studies and surveys, the ethics of AI is discussed in the context of the three stages of AI concepts and their impact on humans. Next week we will provide a technical perspective on privacy and data protection, which is at the top of this list.

References:

  1. 1.
    Shaping the Ethical Dimensions of Smart Information Systems, A European Perspective, https://www.project-sherpa.eu/
  2. 2.
    Bernd Carsten Stahl, "Artificial Intelligence for a Better Future: An Ecosystem Perspective on the Ethics of AI and Emerging Digital Technologies”, Springer, ISBN-978-3-030-69978-9, 2020.