Western artificial intelligence digital ethics regulations: dilemma and road

Author: Wang Peinan (Teacher of the School of Writing and Communication Teaching of Tsinghua University of Humanities)

Source: "Youth Reporter", No. 13, 2022

With the rapid development of artificial intelligence technology and the continuous expansion of application scenarios, artificial intelligence driven by algorithms is increasingly embedded in different fields of human society. Artificial intelligence technology represented by preference push, Q & A robots and face recognition has become an important part of social life. However, compared with the rapid iteration of technology, the ethical regulations for artificial intelligence technology have been in a wide range of controversy for a long time.

Recently, many mainstream media in the United States have reported on the news that Google's engineer Blake Lemoine has been suspended because he claims that his team's artificial intelligence chat robot Lamda has human emotions and consciousness. Although Google immediately clarified that there was no evidence that Lamda had human consciousness, the development of artificial intelligence technology made it more and more deeply embedded in human society. Smart ethics regulations have put forward new challenges.

Artificial intelligence's moral border debate

In fact, this is not the first time that artificial intelligence chat robots have caused controversy. It is not uncommon for artificial intelligence robots to form a value judgment in the learning process. In 2017, Microsoft launched artificial intelligence chat robot Tay in Twitter. It learns self -learning by dialogue with Twitter netizens, and then approaches human expression by imitating the content of other netizens. But in just one day launch, TAY has evolved into a scrupulous, gathered racialist, patriarchal discrimination, and hatred of Jewish robots. Because the situation was on the verge of losing control, Microsoft had to go offline. Coincidentally, LUDA, a girl chat robot launched in South Korea in 2020, was sexually harassed by users' continuous language, so that LUDA quickly learned to discriminate against minority groups, women and disabled people.

In addition to the value judgment, the ethical issues of artificial intelligence in the decision -making process have also attracted widespread attention. The recruitment of multinational companies such as L'Oreal and Coca -Cola has introduced artificial intelligence systems to participate in resume audits and candidate first screening; Amazon has used artificial intelligence in the management of warehouse employees very early, and tracking employees through artificial intelligence automated employees through artificial intelligence automated employees Once the work conditions are found that the warehouse employees have not touched the package for a long time or the processing speed becomes slower, the employee efficiency will be too low. Amazon's artificial intelligence further calculated the work efficiency of all employees based on the algorithm, and "one -click layoffs" for employees who did not meet the production efficiency indicators many times. The disclosure of this incident has also caused strong controversy in the United States.

The founder of computer ethics, Moore, divides the ethical intelligence into four categories. The current level of artificial intelligence is still in the early stage of the "ethical influence of intelligence". judge. However, when artificial intelligence is increasingly involved in the fields of value judgment and social decision -making, and has significant social impact, the vagueness of human -machine margins will cause more severe power and responsibility to attribute unclear problems. Artificial intelligence products have personalized characteristics in terms of morality and emotion, but they are different from real people. So the artificial intelligence of value judgment and social decision -making should only be regarded as the subject of performing tasks, or can it be regarded as a moral subject to take responsibility? The process of artificial intelligence involving human society and generating social consequences has brought re -thinking about individual, technical and moral relations.

Technology and Law: Two dilemma of digital ethics regulation

Although effective regulations on artificial intelligence ethics have gradually become consensus, the reality level regulation still faces the dilemma of technical and legal dimensions. At the technical level, the regulation of artificial intelligence ethics will be caught in the transparent trap brought by technical black boxes. Because the algorithm has the ability to learn and adjust, the learning and output process that developers may not be predicted and explained during the operation of artificial intelligence algorithms may occur. Some scholars believe that when artificial intelligence provides services to the society or makes decisions, it shall give users the right to obtain algorithm explanation. However, at the practical level, artificial intelligence algorithms, as developers' unique intellectual property and the core commercial secrets of Internet enterprises, are almost impossible to regulate through public source code or join the third -party review mechanism.

This has led to the technology gap in the supervision of artificial intelligence decision -making. Social media users cannot explain why pushing some information is prioritized, and Amazon employees cannot explain that they are judged to be inefficient and negatively idle because of their improper behavior. Technical neutrality has become a shame of social embedding that cover technology, but once artificial intelligence replaces social judgments, the ethical justice contained in it will inevitably be questioned. The preferences of accurate push will affect the environment and public issues of public opinion. Enterprises using artificial intelligence to recruit and regulate employee may contain the issue of "quantitative society" that leads the basic rights and interests of workers.

At the legal level, whether artificial intelligence can be used as a moral subject that has independent judgment and can be responsible for its social judgment is a gap in the legal supervision level. The emergence of artificial intelligence has impacted the existing legal order, because traditional legal personality includes the subject of natural persons and the law of law. The biggest difference between the subject of the natural person and the law of the law is that the legal personality accompanies the life of the natural person, and the legal personality of the law must be qualified to enjoy the right and fulfill the obligations after the judgment of the law. Artificial intelligence is neither the main body of nature nor the scope of the existing legal subject, which leads to theoretically no consequences do not need to judge and social decision -making. In the face of such a predicament, some scholars have proposed limited consequences of artificial intelligence to undertake behavior, so they have limited legal personality, and some scholars try to further extend the limited personality theory. Essence However, no matter what kind of perspective, there is no realistic dilemma brought about by eclecticism. In addition to the limited personality of artificial intelligence, who can take responsibility for the realistic problems it causes? For example, Amazon's decision -making power of layoffs to artificial intelligence, is it in line with its limited personality status? Considering that artificial intelligence is embedded in various fields of social structure, a legal regulation principle that consistent with the right to ethical value is needed to get further attention.

Resort of the principle of consensus: ethical regulation of artificial intelligence entering the way

The Montreal Ethics Research Research Director has been focusing on the ethical disputes and governance issues brought by artificial intelligence technology for a long time. The Institute publishes a "Artificial Intelligence Ethics Report" each year to track the ethical regulations used by artificial intelligence. The report of 2022 pointed out that the ethical regulations of artificial intelligence are in parallel in parallel in the "excessive macro principle and lack of operation tools". On the one hand, public and public opinion shows obsession with new technologies, and on the other hand, they have to face the fear brought by technology out of control. The "Artificial Intelligence Ethics Report" believes that compared with the principles of general and macro, responsible artificial intelligence ethics needs to establish a set of specific and clear operability, and include the following basic principles:

(1) Based on human moral framework. Regardless of how artificial intelligence develops at the technical level, it still needs to follow the moral framework and guidelines of human society when it is finally serving human society. The long -term debate on the subjectivity of artificial intelligence and morality may not be helpful for promoting the corresponding governance process, because artificial intelligence does not have the uniqueness and superiority beyond the principle of human morality at the ethical level. Amazon's measures to use artificial intelligence to supervise employees and decide the layoff list cannot be exempted from Amazon's moral responsibility for formulating this rule because of the involvement of artificial intelligence. Therefore, artificial intelligence enterprises and development teams, as technology developers, need to carry out technology development based on basic moral standards, and bear morality and legal responsibilities for their consequences.

(2) Diversity data learning. As mentioned earlier, many artificial intelligence chat robots with independent learning capabilities from different countries research institutions have begun to publish racial discrimination and hatred remarks after a period of "natural" training, which has triggered researchers for artificial intelligence Thinking of the source of data. Artificial intelligence chat robots have accepted a lot of extreme and negative information in social media, so that they unconsciously moved to the problem of violation of social ethics in autonomous learning and training. Therefore Different sources and types of data make the cognition and expression of artificial intelligence closer to the general level of society, rather than some extreme extreme states.

(3) Examination of algorithms in the public sector. As an important governance approach to artificial intelligence algorithms, the government and public sector holding accountability is becoming a trend of digital governance. The EU has passed the "Digital Service Law" in April this year, and puts forward higher requirements for the transparency of the recommendation algorithm. my country also passed the "Guiding Opinions on Strengthening the Comprehensive Governance of Internet Information Service Algorithms" last year, requiring to establish and improve the supervision and management mechanism for artificial intelligence algorithms within three years. The gradual improvement of relevant laws and regulations means that the governance of artificial intelligence issues has gradually embarked on the "fast lane", and the intervention of the public sector has also made the ethical regulations of artificial intelligence a important public issue.

As the direct consequence of technological development, artificial intelligence is becoming an important interest in many areas of human society. Facing the continuous iteration of its technical form, governments, Internet companies and users of various countries need to reach a complete and operable rules, and to achieve the orderly development of artificial intelligence technology on the basis of resorting to consensus.

Reference format for this article:

Wang Peinan. Western artificial intelligence digital ethics rules: Dilemma and Entry [J]. Young reporter, 2022 (13): 95-96.

- END -