The author provides many lists and descriptions of the ethical principles of artificial intelligence and classifies them into eight topic trends. Privacy, responsibility, security and assurance, transparency and interpretability, fairness and non-discrimination, human control over technology, professional responsibility and promotion of human value. (Fjeld and Nagy, 2020)

The Illusion of Ethical Artificial Intelligence

I think there is a special principle in these lists, which is to take care of the non-human world. As Boddington said in her book Ethical Code of Moving toward Artificial Intelligence (2018): "...We are changing the world, and AI will accelerate these changes, so we'd better have a concept of what changes are good and what are bad." (, 2018) We all have different views on this, but this needs to be part of the discussion. We cannot continue to destroy the earth while trying to create super artificial intelligence, and also fantasize that our moral principles are saving the world.

This will also be a cautionary tale, because many of these principles are theoretically reasonable, but they are like a veil, presenting moral illusions. This can be dangerous because it makes us feel like we are practicing ethics while the business is going on as usual. Part of this is because the field of ethical artificial intelligence development is so new that a lot of research has not been conducted to ensure that the overall impact is socially beneficial. "Although these 'AI principles' are emerging, there is little focus on the understanding of these efforts, both individually and in the context of the ever-expanding principle universe, with obvious trends." (Fjeld and Nagy, 2020)

The principle is a double-sided coin. On the one hand, it is good to make clear efforts to follow a range of moral principles. It is good for people to think about how to do the right and ethical things instead of blindly typing codes that can cause damage in unforeseen ways.

Some principles are simple on the surface, but they are incredibly challenging in practice. For example, if we look at the commonly adopted principle of transparency, saying that algorithms and machine learning should be interpretable, and the actual developed method can see the inside of the black box, there is a big difference between this. As the dataset grows larger, this brings more and more technical challenges. (, 2018)

Furthermore, some principles may conflict with each other, which may lead us to a more immoral place than we did when we started. For example, transparency may conflict with another popular principle - privacy. Around this, we may encounter many complex problems, and I hope to see this problem be solved quickly and thoroughly as we move forward.

Overall, we want these concepts to be in people’s minds: like fairness. Accountability, and transparency. These are the core tenets and titles of the FAACT Conference, which explores these principles in depth. It is extremely important for businesses and programmers to focus on topics such as prejudice, discrimination, oppression and systemic violence that are generally involved. However…what may happen is that these principles make us feel like we are doing the right things, yet how many things can actually change when writing these ideals?

The Ethical Revolution of Artificial Intelligence We Need

To make AI ethical, a lot of things have to change, not just in the technology field. People seem to overlook some unknown principles: the value of money for businesses and those in power, and convenience for those with the ability. If we try to create equity, accountability and transparency in AI, we need to do some serious work to society, adjusting our core principles from money and convenience to taking care of everyone’s basic needs and the planet.

Can artificial intelligence become a tool, and its side effect is to launch an ethical revolution?

How do we achieve this? The language we use is important, especially when it comes to principles. Moss and Metcalf pointed out the importance of using market-friendly terms. If we want morality to win, we need to prove the necessary organizational resources, and more often, companies will choose profits over social welfare. (Moss and , 2019)

et al. describe the need to focus on the tensions in ethical in AI and point out the ambiguity of terms such as "equity", "justice" and "autonomy". The authors prompt us to question the possible interpretation of these terms in different groups and contexts. ( et al. 2019)

They continue to say that in order to play a role in practice, principles need to be formalized into standards, codes, and ultimately into regulations. They draw attention to the importance of acknowledging the tensions between high-level goals of morality, which may be different or even conflicting. To be effective, some degree of guidance on how to resolve different situations must be included. To embody a true agreement, different perspectives and values ​​must be recognized and tolerated as much as possible. ( et al. 2019)

The authors then introduce four reasons why tensions are beneficial and important to AI ethics:

Each one needs to be considered continuously, because these tensions cannot be resolved overnight. In particular, it is important to establish a bridge between principles in practice, as I argued above.

As a summary, I will share this direct quote because it is incredibly profound.

"We need to balance the requirement to make our moral reasoning as strong as possible with preventing it from being too rigid and throwing the moral baby away with the bath water by rejecting anything we cannot explain immediately. This has a lot to do with the attempt to draft moral codes and to implement moral reasoning in machines." (, 2018 p.18-19)

In short, ethics, or ethical principles of artificial intelligence, are important, and I like the conversations that start because of their existence. However, it cannot stop there. I'm glad to see more and more ways to put these principles into action and to see technicians and theorists working together to study how to make them work. I also hope we can open our minds and go beyond the idea of ​​making money and creating convenience for businesses, but instead towards resolving tensions and truly creating a world that serves everyone.

refer to

ACM Fairness, Accountability and Transparency Conference (ACM FACCT). ACM FAccT. (2021). January 7, 2022, retrieved

Ai's Principle. of Life. (2021, December 15). Searched on December 30, 2021, from

.(nd). Searched on January 8, 2022, from

,Paula.(2018).gence.lpu.

Fjeld , J., & Nagy, A. (2020). Principled Artificial Intelligence. Klein . December 30, 2021, from.

, F. (2021). on : Four principles of explainable artificial intelligence: 35 comments. . Searched on January 7, 2022, from

Moss , E., & , J. (2019, 14). Ethical dilemma at the core of large tech companies. Harvard Business Review. Searched on December 13, 2021, from.

Ai & Research Institute. (nd). Machine learning principles. 8 principles for responsibly developing artificial intelligence and machine learning systems. Searched on December 30, 2021, from

, J., Cave, S., , A., & Nyrup, R. (2019). The role and limitations of principles in artificial intelligence ethics. a ... Searched on December 13, 2021, from.

This article is an original article from EET Electronic Engineering Album and is prohibited from reprinting. Please respect intellectual property rights, and we reserve the right to hold liable for violators.