AI and Law

The research area ‘AI and Law’ has a long history. Already in the 1940’s academic discussions on technical tools for analysis of legal decisions and regulations by means of technology appeared, but related joint efforts can be traced further back than that. Boole used Jewish law as an example in ‘The Laws on Thoughts’ (1854) and Hollerith’s tabulating machine was constructed in order to count people in a national census prior to presidential elections, as stipulated in legislation (1890). Modern research concerning computer science, including AI and Law, is documented from as early as the 1960s and international conferences on AI and Law have been held on a regular basis since the 1980s. In November 2020 the 35th yearly Nordic Conference on Law and IT ‘Law in the Era of Artificial Intelligence’ is planned to be held in Stockholm.

Research in AI and Law at Stockholm University (SU) encompasses both methodological aspects and regulatory issues. The approach assumes that technology must be designed so that compliance with legal and ethical principles is ensured. It is thereby postulated that AI cannot only reflect technical and operational efficiency, AI must also be politically, socially and ethically acceptable. Nor can systems that are able to change their operations autonomously be controlled solely by written law and legal principles – the regulatory frameworks controlling these systems must be interpreted, recast and embedded in their design. In this respect, AI and Law is a significantly multidisciplinary field of research and legitimate AI presupposes an amalgamation of jurisprudential, social, ethical and technical knowledge.

Methodological oriented research in AI and Law include, inter alia, autonomous vehicles (adhering to traffic legislation), smart buildings, live-in-labs and health care (ensuring privacy by design, accountability and non-discrimination), legislative techniques and automated legal decisions (jurisprudential system analysis and algorithmic development), e-government (embedding administrative laws, freedom of information and secrecy), consequential analysis of facial recognition, deep fakes and machine learning, as well as classification of algorithms and sensitive data sets. In parallel, research concerning regulatory issues include but are not limited to redistributions of liabilities, transparency, privacy, intellectual property rights, data and system security, vulnerability, new forms of criminal activities and development of ethical and legal impact assessment standards.

The ambition is to strengthen and ensure a logical development of an international research environment of excellent quality, able to operate at the forefront of AI and ELSI (Ethical, Legal and Social Impact) research. At SU research in AI and Law is conducted at the world’s first research organisation focusing on the interaction between Law and IT, the Swedish Law and Informatics Research Institute (IRI) ‘irilaw.org’, established in 1968.

More than 50 years of sustained efforts have made it possible to establish a close to unique position in this area of research, nationally and internationally. The co-workers of IRI are experienced as legal and ethical experts in international ICT/AI research projects (FP7, H2020, Marie Skłodowska-Curie) and IRI is founding partner of the national initiative AI Innovation of Sweden (‘ai.se’ launched in 2019). The unit is engaged in training schools and administering secondments for Early Stage Researchers in Marie Curie and COST actions. IRI has also hosted a Master Program in Law and IT for more than ten years, enrolling students from close to 70 countries.

As for novelty, new forms of AI draw the attention to new types of problems in almost all sectors of society – legal, ethical and social consequences are being generated at an unforeseen pace. The illustrations are close to countless. Machine learning provides a new paradigm for efficiency, facial recognition enhance security and deep fakes is a ground-breaking tool for entertainment and the gaming industry, but these and other AI related phenomena may also challenge fundamental ethical and legal principles. The technology may even alter the preconditions for democracy and unpredicted applications as well as unintended usage may lead to manipulation, discrimination and illegitimate surveillance. ELSI of AI is a moving target in critical need of increasing attention.

The need to address the societal challenges AI brings about are broadly recognised. So far however few responses reflect more than tentative attempts to extrapolate methods and juxtapose solutions originating from established sciences – lawyers try to amend static black letter law with lists of vague ethical principles in order to regulate autonomous processes. At the same time computer scientists develop code with little or no interaction with jurisprudential research, often being oblivious of wider societal consequences. In contrast, proficient research in AI and Law presupposes efforts beyond established academic topics and since its inception the exploration of AI and Law has to a large extent been dependent on multidisciplinary work and external financing, often project oriented and of uncertain perseverance.

The increased demand for research and education alter the preconditions. The need for more sustainable efforts is obvious and from the point of view of personnel the situation is challenging.