人工智能的可解释性与AI的法律责任问题研究

人工智能(AI)作为类人类智能,无论我们是否赋予其主体资格,在解决其法律责任问题时,都必须对其行为进行解释,为此,探讨人工智能的法律责任问题,应该基于人工智能行为的可解释性的全新路径来推进,而不是纠缠于当下学界关于人工智能主体地位与法律责任的各种主体论与责任理论。人工智能的可解释性,亦即解释人工智能如何在大数据的基础上进行算法决策。然而,在AI领域,虽然以深度学习为代表的人工智能技术已取得了令人瞩目的成就,但如何确保以非技术性的方式向最终用户和其他利益相关方解释算法决策以及任何驱动这些决策的数据,仍是一个无法得到解决的难题,人工智能"黑箱"释明难题决定了人工智能行为的不可解释性。法律责任的本质是答责,不具有可解释性的人工智能不能自我答责,因此其无法承担法律责任;法律责任的目的是预防,不具有可解释性的人工智能无法实现法律责任的预防目的。人工智能法学研究的下一个前沿问题,是人工智能的可解释性问题。     

Educated guesses: health risk assessment in environmental impact statements

Environmental pollution threatens public health. The search for solutions has advanced the frontiers of science and law. Efforts to protect the environment and public health begin with describing potential adverse consequences of human activities and characterizing the predicted risk. The National Environmental Policy Act requires the preparation of environmental impact statements to describe the effects of proposed federal projects and provide information for agency decisionmakers and the public. Risks to public health are particularly difficult to quantify because of uncertainty about the relation between exposure to environmental contamination and disease. Risk assessment is the current scientific tool to present estimates of risk. The methodology has created controversy, however, when underlying assumptions and uncertainties are not clearly presented. Critics caution that the methodology is vulnerable to bias. This Note evaluates the use of risk assessment in the environmental impact statement process and offers recommendations to ensure informed decisions.     

Using sensitive data to prevent discrimination by artificial intelligence: Does the GDPR need a new exception?

Organisations can use artificial intelligence to make decisions about people for a variety of reasons, for instance, to select the best candidates from many job applications. However, AI systems can have discriminatory effects when used for decision-making. To illustrate, an AI system could reject applications of people with a certain ethnicity, while the organisation did not plan such ethnicity discrimination. But in Europe, an organisation runs into a problem when it wants to assess whether its AI system accidentally discriminates based on ethnicity: the organisation may not know the applicants’ ethnicity. In principle, the GDPR bans the use of certain ‘special categories of data’ (sometimes called ‘sensitive data’), which include data on ethnicity, religion, and sexual preference. The proposal for an AI Act of the European Commission includes a provision that would enable organisations to use special categories of data for auditing their AI systems. This paper asks whether the GDPR's rules on special categories of personal data hinder the prevention of AI-driven discrimination. We argue that the GDPR does prohibit such use of special category data in many circumstances. We also map out the arguments for and against creating an exception to the GDPR's ban on using special categories of personal data, to enable preventing discrimination by AI systems. The paper discusses European law, but the paper can be relevant outside Europe too, as many policymakers in the world grapple with the tension between privacy and non-discrimination policy.     

Sex offender risk assessment: Consideration of human rights in community protection legislation

Objectives. Risk assessments provided to judicial decision makers as a part of the current generation of legislation for protecting the public from sexual offenders can have a profound impact on the rights of individual offenders. This article will identify some of the human rights issues inherent in using the current assessment procedures to formulate and communicate risk as a forensic expert in cases involving civil commitment, preventive detention, extended supervision, or special conditions of parole. Method. Based on the current professional literature and applied experience in legal proceedings under community protection laws in the United States and New Zealand, potential threats to the rights of offenders are identified. Central to these considerations are issues of the accuracy of current risk assessment measures, communicating the findings of risk assessment appropriately to the court, and the availability of competent forensic mental health professionals in carrying out these functions. The role of the forensic expert is discussed in light of the competing demands of protecting individual human rights and community protection. Conclusion. Actuarial risk assessment represents the best practice for informing judicial decision makers in cases involving sex offenders, yet these measures currently demonstrate substantial limitations in predictive accuracy when applied to individual offenders. These limitations must be clearly articulated when reporting risk assessment findings. Sufficient risk assessment expertise should be available to provide a balanced application of community protection laws.     

Can legitimate interest be an appropriate lawful basis for processing Artificial Intelligence training datasets?

Precision and effectiveness of Artificial Intelligence (AI) models are highly dependent on the availability of genuine, relevant, and representative training data. AI systems tested and validated on poor-quality datasets can produce inaccurate, erroneous, skewed, or harmful outcomes (actions, behaviors, or decisions), with far-reaching effects on individuals' rights and freedoms.Appropriate data governance for AI development poses manifold regulatory challenges, especially regarding personal data protection. An area of concern is compliance with rules for lawful collection and processing of personal data, which implies, inter alia, that using databases for AI design and development should be based on a clear and precise legal ground: the prior consent of the data subject or another specific valid legal basis.Faced with this challenge, the European Union's personal data protection legal framework does not provide a preferred, one-size-fits-all answer, and the best option will depend on the circumstances of each case. Although there is no hierarchy among the different legal bases for data processing, in doubtful cases, consent is generally understood by data controllers as a preferred or default choice for lawful data processing. Notwithstanding this perception, obtaining data subjects' consent is not without drawbacks for AI developers or AI-data controllers, as they must meet (and demonstrate) various requirements for the validity of consent. As a result, data subjects' consent could not be a suitable and realistic option to serve AI development purposes. In view of this, it is necessary to explore the possibility of basing this type of personal data processing on lawful grounds other than the data subject's consent, specifically, the legitimate interest of the data controller or third parties. Given its features, legitimate interests could help to meet the challenge of quality, quantity, and relevance of data curation for AI training.The aim of this article is to provide an initial conceptual approach to support the debate about data governance for AI development in the European Union (EU), as well as in non-EU jurisdictions with European-like data protection laws. Based on the rules set by the EU General Data Protection Regulation (GDPR), this paper starts by referring to the relevance of adequate data curation and processing for designing trustworthy AI systems, followed by a legal analysis and conceptualization of some difficulties data controllers face for lawful processing of personal data. After reflecting on the legal standards for obtaining data subject's valid consent, the paper argues that legitimate interests (if certain criteria are met) may better match the purpose of building AI training datasets.     

Promises and limits of law for a human-centric artificial intelligence

While the concept of human-centric artificial intelligence (AI) has emerged as a key principle to govern AI systems, two obstacles for its implementation remain largely understated. First, the excessive focus on accountability at the design stage of AI systems, overshadowing the fact that human values can be affected at different stages across the AI life cycle. Second, the market-driven approach of current regulatory initiatives, limited in their ability to actively promote human values. In this article, we argue for a twofold approach to tackle these limitations. On one hand, we propose a co-evolutionary and life cycle approach to tackle the lack of accountability of AI systems, showing that this approach can help ensure accountability beyond the design stage by enabling meaningful human control and human-AI interaction across the entire lifecycle of the system. On the other hand, we propose that regulatory initiatives should balance the market-driven approach by giving a more predominant role to human rights and by introducing explicitly the notion of proportionality test. This rebalancing would serve to handle conflicts between the objectives pursued by AI systems circulating in the markets and the need for an effective protection of human rights.     

Prediction Versus Management Models Relevant to Risk Assessment: The Importance of Legal Decision-Making Context

Most of the theoretical and empirical literature on violence risk to date has focused on the task of predicting who will behave violently. In the present article, it is argued that at least two models of risk assessment may be applied to the varying legal decisions in which violence risk is a consideration: prediction (with an emphasis on overall accuracy) and management (with an emphasis on risk reduction). These two models are described, and discussed in the contexts of the literatures on forensic assessment and therapeutic jurisprudence. The implications for research, policy, and practice are considered.     

The impact of international human rights norms—a “law undergoing evolution”

Abstract

In this paper, the author outlines the history of, and reasons for, the growing impact of international human rights jurisprudence upon the work of judges in New Zealand, Australia, England and elsewhere in the Commonwealth of Nations. Formerly, international and domestic law were virtually entirely separate. But now, there is increasing legal authority to support the use of international human rights jurisprudence in domestic judicial decision‐making. It can be done in the application of constitutional or statutory provisions reflecting universal principles stated in international treaties. But, according to the Bangalore Principles, it can also be done where there is a gap in the common law or where a local statute is ambiguous. The judge may then fill the gap or resolve the ambiguity by reference to international human rights jurisprudence which will ensure that domestic law conforms, as far as possible, to such principles.

In its decision in Tavita, the New Zealand Court of Appeal declared this to be “a law … undergoing evolution”. The author outlines some of the impediments and problems for the evolution. But he also collects the reasons why it is a natural and inevitable phase of the common law in the current age. He suggests that judges should be aware of the developments. In appropriate cases, they should inform their decisions with relevant international human rights jurisprudence. That will at least ensure that they develop domestic human rights law in a principled way, consistently with international law, and not in an idiosyncratic fashion “discovering” new fundamental rights which may otherwise be criticised as mere judicial invention.     

Indeterminacy,Disagreement and the Human Rights Act: An Empirical Study of Litigation in the UK House of Lords and Supreme Court 1997–2017

This article explores the impact of the Human Rights Act 1998 on the decision making of the House of Lords (UKHL) and the UK Supreme Court (UKSC). How does Convention rights content vary across areas of law in the UKHL/UKSC? Are some judges more likely than others to engage in Convention rights discourse? Is judicial disagreement more common in cases with higher levels of Convention rights discourse? A robust method of answering questions of this nature is developed and applied to decisions of the UKHL/UKSC, showing that the Convention rights content of decisions has varied over time and over substantive areas of law. Higher levels of human rights discourse are associated with greater levels of disagreement. A benchmarked measure of human rights content is developed to show the effect of the particular judge on the human rights content, illustrating the indeterminacy in human rights discourse and how its deployment can be contingent on judicial attitudes.     

Towards a more proactive approach to brand protection: development of the Organisational Risk Assessment for Product Counterfeiting (ORAPC)

Through an adaptation of a terrorism risk assessment model, this article develops an initial proactive product counterfeiting risk assessment that is designed to focus upon a specific product’s risk for being counterfeited. The goal of developing this risk assessment is to help corporations identify the products that are most at risk for counterfeiting, thereby giving them the ability to focus their resources in the areas where the greatest opportunities for crime are present. This risk assessment is intended to serve as the first line of defence in a comprehensive and proactive brand owner strategy centred on identifying product-specific counterfeiting risk. The assessment comprises three factors that, together, capture a product’s counterfeiting risk level: the threat of product counterfeiting, the brand owner’s vulnerability to product counterfeiting and the potential consequences of a counterfeit product entering the market and reaching consumers.     

Dissolving privacy,one merger at a time: Competition,data and third party tracking

Amid growing concern about the use and abuse of personal data over the last decade, there is an emerging suggestion that regulators may need to turn their attention towards the concentrations of power deriving from large-scale data accumulation. No longer the preserve of data protection or privacy law, personal data is receiving attention within competition and antitrust law. Recent mergers and acquisitions between large digital technology platforms have raised important questions about how these different areas intersect and how they can complement one another in order to protect consumer welfare while ensuring competitive markets.This paper draws attention to one particularly complicated kind of digital data-intensive industry: that of third party tracking, in which a firm does not (only or primarily) collect and process personal data of its own customers or users, but rather data from the users of other ‘first party’ services. Mergers and acquisitions between firms active in the third party tracking industry raise unique challenges for privacy and fundamental rights which are often missed in regulatory decisions and academic discussions of data and market concentration. In this paper, we combine empirical and normative insights to shed light on the role of competition regulators in addressing the specific challenges of mergers and acquisitions in the third party tracking industry. After critically assessing some of the US and EU case law in this area, we argue that a bolder approach is needed; one that engages in a pluralist analysis of economic and noneconomic concerns about concentrations of power and control over data.     

Principles and business processes for responsible AI

The first article in this series examined why the world wants controls over Artificial Intelligence (AI). This second article discusses how an organisation can manage AI responsibly, in order to protect its own interests, but also those of its stakeholders and society as a whole. A limited amount of guidance is provided by ethical analysis. A much more effective approach is to apply adapted forms of the established techniques of risk assessment and risk management. Critically, risk assessment needs to be undertaken not only with the organisation's own interests in focus, but also from the perspectives of other stakeholders. To underpin this new form of business process, a set of Principles for Responsible AI is presented, consolidating proposals put forward by a diverse collection of 30 organisations.     

On defense of “ethification” of law: How ethics may improve compliance with the EU digital laws

In recent years, academics and professionals witness the rise of the “ethification” of law, specifically in the area of ICT law. Ethification shall be understood as a proliferation of moral principles and moral values in the legal discourse within the areas of research, innovation governance, or directly enforceable rules in the industry. Although the ethical considerations may seem distant from mere regulatory compliance, the opposite is true. The article focuses on the positive side of the “ethification” of digital laws through the lens of legal requirements for impact assessments pursuant to General Data Protection Regulation and conformity assessments in the proposal for the Artificial Intelligence Act. Authors argue that ethical considerations are often absent in the context of using new technologies including artificial intelligence, yet they may provide additional value for organizations and society as a whole. Additionally, carrying out ethics-based assessments is already in line with existing regulatory requirements in the fields of data protection law and proposed EU AI regulation. These arguments are reflected in the context of facial recognition technology, where both data protection impact assessment under the EU General Data Protection Regulation and conformity assessment under the proposal of the EU Artificial Intelligence Act will be mandatory. Facial recognition technology is analyzed through the ethics-based assessment involving stakeholder analysis, data flows map, and identification of risks and respective countermeasures to show additional insights that ethics provides beyond regulatory requirements.     

Democratic Inclusion,Law, and Causes

Abstract. In this article two conceptions of what it means to say that all affected persons should be granted the right to vote in democratic elections are distinguished and evaluated. It is argued that understanding “affected” in legal terms, as referring to the circle of people bound by political decisions, has many advantages compared to the view referring to everyone affected in mere causal terms. The importance of jurisdictions in deciding rights to democratic influence should hence be recognized more clearly than it currently is in democratic thinking.     

Access to justice: a new approach using human rights standards

This article details a trial of a new approach to measuring access to justice that utilises human rights instruments as the reference point. It involves an examination of people's actual experience of the justice system using human rights standards as the benchmark. The research project selected the right to income security. The project trialled a range of methods gathering data about how people have been treated in the Australian social security system and how they would expect to be treated if there was a human right to social security in Australia. This data is assessed against the set of standards developed to measure the enjoyment of the right to social security. The trial suggests that without knowledge about human rights and legal rights, without the confidence to exercise those rights and without the capacity or capability to seek or find help it is unlikely that people will realise their rights and accordingly access to justice is placed in question. The research methodology has the potential to be a useful model to conduct further access to justice research.     

Sovereignty, Migration and the Rule of Law in Global Times

This article argues that in the present era of globalisation, control over the movement of people has become the last bastion of sovereignty. This is important both to theoretical accounts of globalisation and to policy decisions by governments. Nation states threatened with loss of control in other realms are implementing a variety of 'crackdown' measures in questions of immigration. Issues of refugee law, illegal migration and skilled migration each challenge sovereignty in specific ways. While international human rights standards have made few inroads in questions of migration, recent decisions in England and Australia suggest that the rule of law may be emerging as a counter to traditional executive free reign in matters of migration law.     

森林资源生态效益补偿制度简论

随着人类与自然之间伦理关系的变革 ,承认自然具有内在价值及权利的观点已被接受。具体到森林资源生态效益补偿制度而言 ,其理论依据为 :生态安全原则 ;权利之间协调发展原则 ;自然权利原则 ;生态正义原则。至于具体的森林资源生态效益补偿制度的资金筹措方式 ,可采用国家财政拨款、征收生态安全保险金等多种渠道     

Interpreting the Political Theory in the Practice of Human Rights

In this discussion of The Heart of Human Rights, I support Allen Buchanan’s pursuit of a theory-in-practice methodology for interpreting the foundations and meaning of international legal human rights from within the practice. Following my use of that methodology, I recharacterize the theory of rights revealed by this methodology as political not moral. I clarify the import of this interpretation of international legal human rights for two problems that trouble Buchanan: (1) whether the scope of ‘basic equal status’ is a global or an ‘intrasocial’ standard and (2) whether there is a ‘proliferation’ of rights that risks undermining the legitimacy of international legal human rights. I argue that the scope of basic equal status is global and that the practice of making what he calls ‘new’ rights claims is part of the practice of human rights.