AI-Ethics by Design. Evaluating Public Perception on the Importance of Ethical Design Principles of AI.
https://arxiv.org/pdf/2106.00326.pdf
This article mainly discusses the public perception and importance of AI ethical design principles. Through a large survey study, the researchers asked the German public about the relative importance of AI ethical principles (interpretability, fairness, safety, accountability, accuracy, privacy, machine autonomy). The results show that the German public generally views these moral principles as equally important, with no clear preference. However, further cluster analysis showed that there are different preference models in the German population, with significant differences between these groups in the level of preferred attributes and attribute importance. The article also discusses the composition of these groups in terms of sociology and perspectives on AI, as well as design challenges and social implications. While ethical design is essential for AI systems from a technical perspective, political and legal action is also needed to ensure the implementation of ethical AI.
The article mentions differences in the importance placed on ethical principles by age and occupation. In the study, through a survey of the German public, it is found that a part of the population has higher requirements for all ethical principles for ethically designed AI systems, and this part of the population is usually younger, more educated, and has a higher interest in and trust in AI systems. In addition, we also found that people who are more concerned about safety and people who are less concerned about ethical design. These different populations have different needs and expectations for the ethical design of AI systems, so the needs of different populations need to be considered when designing and implementing AI systems technologically. In addition, the article mentions that the public’s views on the development of AI may influence system design, for example, some people may prefer simple and understandable algorithms over more complex ones, even if complex algorithms perform better in terms of fairness. Therefore, the public’s views on AI development need to receive more attention, but should not dominate other views on AI development and implementation.
I will refer to the importance that people of different identities attach to the moral code in the article to construct the impact of AI taxation on different groups in my paper.
