Periodic Reporting for period 2 - CompuLaw (Computable Law)
Okres sprawozdawczy: 2021-05-01 do 2022-10-31
What is the problem/issue being addressed?
The project addresses the regulation of computations (processes and systems), focusing on Artificial Intelligence through an innovative legal and technological framework. It aims to provide epistemic, technical and normative guidance for the development of computable laws and law-compliant computations.
Technologies based on AI can allow humans to face the grand challenges of humanity, such as maintaining a healthy environment, providing the resources for a growing population (including energy, food, and water), overcoming disease, vastly extending human longevity, and eliminating poverty.
However, the development of AI and its convergence with big data also leads to serious risks for individuals and for the whole of society. AI-based technologies may increase opportunities for control, manipulation, and discrimination; disrupt social interactions, and expose humans to harm resulting from technological failures or disregard for individual rights and social values.
Why is the issue important for society?
Unless computational entities are well-behaved —comply with legal norms, avoid harm and discrimination, respect rights and justified expectations, and transparently explain their behaviour to the individuals concerned— the infosphere may become an unfriendly and dehumanising environment, where individuals are subject to surveillance and to determinations over which they have no control, being outsmarted, cheated, and possibly harmed by faulty, illegal or unethical computations.
Moreover, the scale, speed, ubiquity and autonomy of computations make it impossible for humans to directly monitor them and anticipate all possible illegal computational behaviour.
The law can hold the hybrid infosphere under its rule – providing protection, security and trust – only if it becomes computable, i.e. it must be an internal component of computational processes rather than only an external constraint over them. Accordingly, legal norms, values and principles must be mapped onto, and partially translated into, (a) computable representations of legal information and reasoning, that (b) are processed by computational entities.
What are the overall objectives?
The current legal culture still has not adequately addressed the risks and potentials of computable law. CompuLaw project will fill this gap, providing concepts, principles, methods and techniques, and normative guidelines to support law-abiding computations.
To this end, CompuLaw will provide major methodological and substantive breakthroughs. On the one hand, it proposes a socio-technical methodology for regulatory design and evaluation. On the other hand, it develops a framework including (a) norms, legal values and principles for developers, deployers and users; (b) languages and methods to specify requirements of computations and norms directed to them; (c) cognitive architectures for legally-responsive computational agents.
CompuLaw also supports human empowerment towards and through AI technologies, by enabling individuals and civil society to monitor legal compliance of AI systems, using such systems themselves as affordances for the rule of law.
The project addresses the regulation of computations (processes and systems), focusing on Artificial Intelligence through an innovative legal and technological framework. It aims to provide epistemic, technical and normative guidance for the development of computable laws and law-compliant computations.
Technologies based on AI can allow humans to face the grand challenges of humanity, such as maintaining a healthy environment, providing the resources for a growing population (including energy, food, and water), overcoming disease, vastly extending human longevity, and eliminating poverty.
However, the development of AI and its convergence with big data also leads to serious risks for individuals and for the whole of society. AI-based technologies may increase opportunities for control, manipulation, and discrimination; disrupt social interactions, and expose humans to harm resulting from technological failures or disregard for individual rights and social values.
Why is the issue important for society?
Unless computational entities are well-behaved —comply with legal norms, avoid harm and discrimination, respect rights and justified expectations, and transparently explain their behaviour to the individuals concerned— the infosphere may become an unfriendly and dehumanising environment, where individuals are subject to surveillance and to determinations over which they have no control, being outsmarted, cheated, and possibly harmed by faulty, illegal or unethical computations.
Moreover, the scale, speed, ubiquity and autonomy of computations make it impossible for humans to directly monitor them and anticipate all possible illegal computational behaviour.
The law can hold the hybrid infosphere under its rule – providing protection, security and trust – only if it becomes computable, i.e. it must be an internal component of computational processes rather than only an external constraint over them. Accordingly, legal norms, values and principles must be mapped onto, and partially translated into, (a) computable representations of legal information and reasoning, that (b) are processed by computational entities.
What are the overall objectives?
The current legal culture still has not adequately addressed the risks and potentials of computable law. CompuLaw project will fill this gap, providing concepts, principles, methods and techniques, and normative guidelines to support law-abiding computations.
To this end, CompuLaw will provide major methodological and substantive breakthroughs. On the one hand, it proposes a socio-technical methodology for regulatory design and evaluation. On the other hand, it develops a framework including (a) norms, legal values and principles for developers, deployers and users; (b) languages and methods to specify requirements of computations and norms directed to them; (c) cognitive architectures for legally-responsive computational agents.
CompuLaw also supports human empowerment towards and through AI technologies, by enabling individuals and civil society to monitor legal compliance of AI systems, using such systems themselves as affordances for the rule of law.
Subproject 1, Regulatory issues and frameworks in computable law, focuses on legal issues and the governance of computations. CompuLaw has critically considered the following:
1. The ways in which AI affects legally relevant interests. In particular, we have addressed Human and Fundamental Rights, data protection and consumer protection.
2. Influence chains and legal responsibilities linking computations to their deployers and designers. In particular, we have examined domains in which AI is used to support human decisions, such as in the health and consumer markets
3. The extent to which harm can be prevented and remedied and computational justice and fairness be supported. In particular, we have addressed discrimination toward individuals and groups, by adopting a general as well as a domain-oriented perspective.
Subproject 2, Computational methods and technologies, focuses on the computational modelling of normative requirements and on the implementation of artificial legal agents CompuLaw has addressed the following:
1. Formal representation of computable laws. In particular, we have addressed logical methods for representing laws, burdens of proof, and deontic logic.
2. Intelligent application of legal rules according to precedent cases and legal values/principles. In particular, we have addressed legal interpretation.
3. The use of machine learning methods in the law. In particular, we have addressed the use of deep learning techniques to detect and explain unfairness in contracts, and to model the ethical behaviour of autonomous vehicles.
Subproject 3, A legal-computational approach for the hybrid infosphere, merges law and technology into an integrated framework. The following tasks have been initiated:
1. A prototype online system to support legal argumentation (https://site.unibo.it/compulaw/en/tools)
2. Multiple algorithms for machine learning have been tested for their application in the legal domain
3. The Claudette system (for the detection of unfair clauses in online contracts and privacy policies) is being further developed to fit COMPULAW goals (http://www.claudette.eu/about/index.html)
1. The ways in which AI affects legally relevant interests. In particular, we have addressed Human and Fundamental Rights, data protection and consumer protection.
2. Influence chains and legal responsibilities linking computations to their deployers and designers. In particular, we have examined domains in which AI is used to support human decisions, such as in the health and consumer markets
3. The extent to which harm can be prevented and remedied and computational justice and fairness be supported. In particular, we have addressed discrimination toward individuals and groups, by adopting a general as well as a domain-oriented perspective.
Subproject 2, Computational methods and technologies, focuses on the computational modelling of normative requirements and on the implementation of artificial legal agents CompuLaw has addressed the following:
1. Formal representation of computable laws. In particular, we have addressed logical methods for representing laws, burdens of proof, and deontic logic.
2. Intelligent application of legal rules according to precedent cases and legal values/principles. In particular, we have addressed legal interpretation.
3. The use of machine learning methods in the law. In particular, we have addressed the use of deep learning techniques to detect and explain unfairness in contracts, and to model the ethical behaviour of autonomous vehicles.
Subproject 3, A legal-computational approach for the hybrid infosphere, merges law and technology into an integrated framework. The following tasks have been initiated:
1. A prototype online system to support legal argumentation (https://site.unibo.it/compulaw/en/tools)
2. Multiple algorithms for machine learning have been tested for their application in the legal domain
3. The Claudette system (for the detection of unfair clauses in online contracts and privacy policies) is being further developed to fit COMPULAW goals (http://www.claudette.eu/about/index.html)
COMPULAW will provide major breakthroughs relatively to the state of the art in each of the
mentioned subprojects, building on its interdisciplinary nature.
(a) The project will exemplify a new socio-technical methodology for law & technology studies, where legal and computable norms, as components of sociotechnical systems, can be studied using an engineering design paradigm, profiting of methods of system engineering, socio-technical design and mechanism design. Thanks to deeper understanding of structures of causality, dependency and influence between humans and computations, the project will enable fresh analyses of liabilities and entitlements resulting from computations as well as enhanced models of compliance-by design, including the translation of legal norms into computational requirements for legally-responsive autonomous legal agents (ALAs).
(b) The project will advance logical models of norms, normative systems, normative agents and institutions, by combining formal modelling with a focus on legal issues, legal theories and technological implementations and by providing innovative logical forms, inference schemes, and agent architectures, and techniques for the normative governance of knowledge-based and data (machine learning)-driven AI systems.
(c) The project will provide the first integrated legal & computational regulatory framework for computable law, meant to preserve the rule of law in the hybrid infosphere. The framework will include requirements and standards for designers and deployers, computable norms, and ALAs engineered for compliance. By injecting legal, philosophical and logical methods and ideas into the design of computation-based sociotechnical systems, the project will contribute not only to the law but also to computing. Prototype autonomous legal agents will be implemented in the domains of autonomous vehicles and consumer markets.
mentioned subprojects, building on its interdisciplinary nature.
(a) The project will exemplify a new socio-technical methodology for law & technology studies, where legal and computable norms, as components of sociotechnical systems, can be studied using an engineering design paradigm, profiting of methods of system engineering, socio-technical design and mechanism design. Thanks to deeper understanding of structures of causality, dependency and influence between humans and computations, the project will enable fresh analyses of liabilities and entitlements resulting from computations as well as enhanced models of compliance-by design, including the translation of legal norms into computational requirements for legally-responsive autonomous legal agents (ALAs).
(b) The project will advance logical models of norms, normative systems, normative agents and institutions, by combining formal modelling with a focus on legal issues, legal theories and technological implementations and by providing innovative logical forms, inference schemes, and agent architectures, and techniques for the normative governance of knowledge-based and data (machine learning)-driven AI systems.
(c) The project will provide the first integrated legal & computational regulatory framework for computable law, meant to preserve the rule of law in the hybrid infosphere. The framework will include requirements and standards for designers and deployers, computable norms, and ALAs engineered for compliance. By injecting legal, philosophical and logical methods and ideas into the design of computation-based sociotechnical systems, the project will contribute not only to the law but also to computing. Prototype autonomous legal agents will be implemented in the domains of autonomous vehicles and consumer markets.