Text|Blackstone talks
Editor|Blackstone talks
In order to study the legal issues related to AI technology, it is essential to understand how it operates, especially the logic applied from a legal perspective.
The logic of the application of artificial intelligence technology should include three aspects: the "micro" power embodied in the application of artificial intelligence technology, which needs to rely on the practice mode of traditional public power and the concept of operativism that guides practice.
First, the operation of the "micro" power of artificial intelligence technology application
(1) Emphasize productive "micro" power
"Technology" is a system of ways and methods to control and transform the world through the interaction of human knowledge and wisdom with objective means (software and hardware), and is a body of knowledge about "how to do it".
Therefore, technology is actually a manifestation of knowledge.
Nietzsche
Nietzsche linked the seemingly unrelated concepts of knowledge and power and concluded that "knowledge is power."
Foucault further expands this view, arguing that knowledge and power "coexist" in modern society and become a "knowledge-power" combination.
Moreover, the operation of the "knowledge and power" combination in modern society has been institutionalized, forming a tangible and powerful truth system.
Foucault
Technology based on "knowledge power" exhibits a kind of "micro" power.
This power differs from traditional power in that it emphasizes the productive nature of power rather than the repressive and distorting effect; It is not concerned with beliefs, but with the participation of power in action and practice.
Thus, Foucault argues that this power must be analyzed as something circular, or rather, as something that can only function in the form of chains.
It has never been confined here or there, never in anyone's hands, never as a commodity or wealth, but used and exercised through a web organization.
This decentralized "micro" power permeates all corners of society and individuals, and people, society, and others have become the object of analysis of power mechanisms and are treated as a problem to be solved.
In this process, power begins to produce knowledge and empowers knowledge in the process of generation.
A classic example is science, another manifestation of knowledge, which forces people to say certain things through facilities such as laboratories that risk being considered fraudulent.
Therefore, Foucault believed that power and knowledge are directly linked to each other, and that there can be no power relations without corresponding construction of a field of knowledge, and there will be no knowledge without presupposing and constructing power relations at the same time.
As a form of knowledge, artificial intelligence technology and power are of course also linked, so the process of artificial intelligence technology application is also the operation process of "micro" power.
(2) "Micro" power with the ability to discipline
"Micro" power, although it emphasizes productivity, can eventually be empowered even by science.
While analyzing and solving problems, it is also disciplining people, and this power that seems to be in the hands of no one is the covert exercise of power control by emphasizing technological productivity.
Since the birth of artificial intelligence technology, after years of development, it has been able to make decisions comparable to or even stronger than humans in many fields.
For example, in a competition to predict whether a financial ombudsman would allow a claim, an AI system and a hundred human lawyers each obtained several basic facts about misselling cases, but in the end, the AI was 88.6% correct, compared to 66.3% for human lawyers.
On the one hand, this shows that artificial intelligence technology has been relatively mature in the project.
But on the other hand, it is worrying that if artificial intelligence is used to assist human lawyers, human lawyers may lose their judgment due to too much trust in the judgment of artificial intelligence, and eventually fall into the dilemma of being controlled by artificial intelligence technology.
This control is obviously not a "king's order" or a legal compulsion, but is achieved through the productive nature of artificial intelligence technology.
Therefore, artificial intelligence technology will reflect the discipline ability of "micro" power in the application process, which makes the technology difficult to control and shows the ability to covertly control everyone.
2. Practices attached to traditional public power
Generally, we believe that the era of artificial intelligence is the era of the Internet of Everything, the era of de-traditional centralization and centralization, and the trend of private subjectivization and decentralization of traditional state monopoly power is irreversible.
But in fact, the Internet has evolved from the "decentralization" of the 1.0 era to the "recentralization" of the network platform in the 2.0 era.
"Decentralization" makes the power of a single subject of the state decentralized, but "recentralization" does not concentrate traditional public power in a single subject, but in the hands of the state and multiple social subjects in the context of "decentralization".
These phenomena actually reflect the inseparable relationship between artificial intelligence technology and traditional public power.
The operation of "micro" power embodied in the application of artificial intelligence technology does not seem to show the directional and controlling power of traditional public power, but the cases of technology infringing on individual rights in practice show the pointing and controlling power of power.
This contradiction arises because technology itself does not exist alone, and the subject who masters technology (which is usually the holder of traditional public power) often achieves the directional control effect of traditional public power through technical means.
In other words, this shows that technology shows its orientation and control in practice by being attached to traditional public power.
Among them, traditional public power can actually refer to social public power in addition to the public power of the state.
Social public power is similar to state public power in that its source is not based on violence, but on contract basis.
Power arises from the consent of the subject of the contract, and the coercive and dominant power of the power comes from the agreement (law) of the subject of the contract.
Therefore, social public power is still a traditional public power, which is the dominance of social resources (material and spiritual resources) owned by social subjects (citizens, especially social groups, non-governmental organizations) over society and the state.
Today, the huge impact of large social entities such as Tencent, Sina, and Alibaba on society through technical means (social resources) is the embodiment of social public power.
The holders of traditional public power use technology to achieve control over multiple subjects from both the state and society through the combination of traditional public power.
The same is true in artificial intelligence technology, new data collection methods (including facial recognition, fingerprint recognition, two-dimensional code, etc.) in all corners of society can provide enough data for the analysis of big data, through the analysis of data by artificial intelligence technology, traditional public power holders will hold information that other subjects do not have, and the resulting information asymmetry will make these subjects achieve unprecedented control over the entire society through the application of artificial intelligence technology.
Since the application of artificial intelligence technology reflects the operation process of "micro" power, the holders of traditional public power can use the characteristics of "micro" power to show the specific directional control of traditional public power more hidden.
This is because the characteristics of "micro" power, which are not in anyone's hands and the characteristics of emphasizing production, provide a natural umbrella for the directivity and control of traditional public power attached to artificial intelligence technology, and hide the characteristics of traditional public power behind it.
This allows the holders of traditional public power to blind other subjects, especially individuals, with the shield of technological neutrality, so that they are controlled without knowing it.
Therefore, when artificial intelligence technology is attached to the application of traditional public power, it emphasizes its productivity for all subjects, and on the other hand, in the name of emphasizing productivity, the combination of technology and traditional public power exerts hidden control over specific subjects.
Third, the guiding concept of operivism
In the application practice of technology, technocracy is a common guiding concept for technology application.
Technocracy in a general sense refers to the form of society in which industrial society reaches its peak of organized integration.
Habermas elaborated more on technocracy, which refers to the fact that science and technology began to regulate economic, political, cultural and social life, affecting human production methods, lifestyles and ways of thinking, and becoming a force that dominated the world.
Artificial intelligence technology, which is attached to traditional public power, is guided by the operability in technocracy in practice.
As an externalization of technocracy, operationalism equates the meaning of concepts with the corresponding set of operations, i.e. treating the names of things as representations of their modes of action at the same time, and the names of properties and processes as symbols of those instruments used to perceive or produce them.
The former can only have the content of what the word refers to in the use of propaganda and standardization, and the reaction that people expect the word to provoke can only be the act (reaction) that is promoted and standardized.
This would make a concept that originally contained multiple meanings be equated with a certain meaning within the concept, that is, the ambiguous concept guided by operationalism would be reduced to a specific referent.
The Leibniz paradox believed that all human thoughts could be expressed through a universal language that could function mathematically precisely, and therefore, social problems could be solved by computational means.
For example, in the legal field, under the guidance of operivism, people decompose the law with ambiguous concepts and even require the judge's free mental evidence into one specific referential program after another, and then enter it into the machine, and the machine makes a judgment according to the conditions of the case, so as to achieve the purpose of replacing the judge's decision, which is the representative of operivism one by one famous vending machine judge.
However, judges embodying this ultimate form of rationality should be denied because they only have the efficiency, accuracy and certainty of vending machines and the mechanical operation of legal concepts, have no ethical, social and utilitarian considerations, and ignore the dynamic application of the law by specific legal practitioners in specific cases.
In the initial stages of artificial intelligence, machines were programmed to make a large list of detailed rules to try to replicate human thought in a "what if" fashion, which is the Weber-style vending judge.
Today, artificial intelligence has become a more effective tool after the development of algorithms that teach machine learning, an idea that has developed from the study of pattern recognition and learning theory, and machine learning has become one of the main ways to achieve artificial intelligence.
The current deep learning approach is the main way machine learning achieves artificial intelligence, which gains greater power and flexibility by presenting the world as a nested hierarchy of concepts, each defined according to simpler concepts and more abstract expressions calculated according to less abstract concepts.
That is, it can enable computer systems to mimic human neural networks for more efficient machine learning.
Traditional linear machine learning algorithms differ from deep learning algorithms, and deep learning algorithms are a structure that increases complexity and abstraction.
Therefore, the way of deep learning can make machines learn faster and more accurately.
Because of this, new vending machine judges, such as legal information retrieval systems and the new legal expert system (chatgpt), quickly appeared and became popular.
But also as mentioned above, its intelligence is achieved by machine learning using deep learning algorithms, and its lower level is composed of many simple concepts, and the multi-level nested hierarchy is essentially nothing more than reducing the externalized operivism of technocracy from the past to specific, simple, single signifier to today's specific, complex, several signifiers, but can not restore the entire "concept".
Thus, this set of finite rules is nothing more than a cumulative accumulation of completed human experience, but with a new type of processing.
More comprehensively documented, this is still nothing more than a bundle of material conditions that humans have created.
This artificial intelligence is moving towards an infinite approach to the concept itself, and the flaws of the ultimate formal rationality brought about by its operationalism have been smoothly masked by this trend.
Resources
1. Ye Liangfang and Ma Luyao: "Criminal Law Response to Artificial Intelligence Crime under the Risk Society Viewing Valve", Zhejiang Academic Journal, No. 6, 2018.
2. Ye Qiang: "A Review of German Autonomous Driving Legislation", in Foreign Social Sciences, No. 2, 2022, 88.Yu Haiphong: "The Value Orientation and Logical Premise of Artificial Intelligence Legal Regulation I-Between Replacing Humans and Enhancing Humans", in Law Science, No. 6, 2019.
3. Yu Mouchang: "Out of Anthropocentrism", in Studies in Dialectics of Nature, No. 7, 1994, 90.Zhan Fuman, Miao Jing: "Modern Interpretation of Limited Government Theory", Legal Science (Journal of Northwest University of Political Science and Law), No. 3, 2005.
4. Zhang Aijun and Li Yuan, "Algorithmic Power in the Age of Artificial Intelligence: Logic, Risk and Regulation", Journal of Hohai University (Philosophy and Social Sciences Edition), No. 6, 2019.
5. Zhang Fuli and Zheng Haishan, "The Positioning, Prospect and Risk Prevention and Control of Artificial Intelligence Assisted Sentencing in the Era of Big Data", Guangxi Social Sciences, No. 1, 2019.
6. Zhang Fuli, "The Governance of Artificial Intelligence in a Global Risk Society: A Complexity Paradigm and Legal Response", Academic Forum, No. 3, 2019.