An approach to some eu initiatives on the regulation of liability for damage caused by ai-systems
| Autor | Miquel Martín-Casals |
| Cargo | Full Professor of Private Law, Institute of European and Comparative Private Law, University of Girona, Spain |
| Páginas | 3-24 |
Revista Ius et Prax is, Año 28, Nº 2, 2022
Miquel Martín-Casals
pp. 3 - 24
3
Revista Ius et Praxis
Talca, Chile, 2022
Artículo
AN APPROACH TO SOME EU INITIATIVES ON THE REGULATION OF
LIABILITY FOR DAMAGE CAUSED BY AI-SYSTEMS
Una aproximación a algunas iniciativas de
la UE sobre la regulación de la
responsabilidad por daños causados por
Sistemas de IA
MIQUEL MARTÍN-CASALS*
University of Girona
Abstract
This paper deals with three of the current endeavours of the European Union regarding the regulation of
artificial intelligence (AI). First, it briefly a nalyses the Draft Prop osal of the “Artificial Intelli gence Act” (AI Act),
which puts fundamental rights at the core of Europe’s AI approach and establishes a risk regulation system of
preventive character that proposes ex ante solutions but does not provide for liability rules. Then it refers
briefly to the Resolution of European Parliament adopted in October 2020 which formulated
recommendations to t he Commission for the regulation of civi l liability regardin g the use of artificia l
intelligence, which included the full text of a Draft Regulation “on liability f or the operation o f artificial
intelligence-systems”. Finally, the paper deals in more detail with the ongoing debate on how to adapt the
Directive 85/374/EEC o n liability for defective products to overcome the problems posed by AI, such as the
openness of AI-systems, and other characteristic s of AI that make it difficult to id entify the potentially liable
person, to prove the defect of a product, to prove causation and to apply other rules included in the Directive
now in force.
Keywords
Challenges of AI-s ystems, risk regulation approa ch versus liability rules for dama ge
caused by AI-systems, adaptation of the Dir ective 85/374/EEC on liability for
defective products.
Resumen
Este trabajo trata tres iniciativas actuales de la Unión Europea (UE) para regula r la inteligencia artificial (IA).
En primer lugar, analiza brevemente el Anteproyecto de “Ley de Inteligencia Artificial”, que sitúa los derechos
fundamentales en el centro del enfoque europeo de IA y establece un sist ema de regulación de riesgos de
carácter preventivo que propone soluciones ex ante pero no contempla reglas de responsabilidad civil.
También brevement e, se refiere a la Resolución del Pa rlamento Europeo adoptada en octubre de 2020 que
formuló recomendaciones a la Comisión para la regulación de la responsabilid ad civil en relación con el uso
de inteligencia artificial, que incluía el texto íntegro de un Proyecto de Reglamento “sobre responsabilidad
por el funcionamiento d e sistemas de inteligencia artif icial”. Finalmente, el trabajo trata con más detalle el
debate en curso sobre cómo adaptar la Directiva 85/374/CEE sobre responsabilidad por productos
defectuosos para superar los problemas que plantea la IA, como la apertura de los sistemas de IA y otras
características que dificultan la identificación del posible responsable, la prueba del defecto de un producto,
la prueba de la rela ción de causalidad y la a plicación de otras normas incluidas en la Directiva actu almente en
vigor.
Palabras clave
Retos de los sistemas de IA, regulación de riesgos versus reglas de responsabilidad por daños causados por sistemas
de IA, adaptación de la Directiva 85/374/CEE sobre responsabilidad por productos defectuosos.
* Full Professor of Private Law, Institute of European and Com parative Private Law, University of Girona, Spain, orcid.org/0000-0003-
1743-780X, e-mail: miquel.martin@udg.edu.
Revista Ius et Prax is, Año 28, Nº 2, 2022
Miquel Martín-Casals
pp. 3 - 24
4
1. Introduction
1.1.The fourth industrial revolution and AI as a disruptive technology
Over the past two-hundred seventy years technological development has changed not
only the way human beings organise their work to produce goods and render services, but it has
also transformed political systems and social and legal institutions in a series of what has been
called ‘industrial revolutions’1.Thus, the first industrial revolution spanned from about 1760 to
around 1840 on the wake of the invention of the steam engine which enabled the mechanical
production in factories powered by steam boilers and the construction of railways which created
entirely new systems of communication, exchange, and distribution. A new wave of interrelated
technologies in the period between 1870 and 1930 gave rise to a second industrial revolution
which increased the growth and opportunities that came from the first one. The advent of the
assembly line made mass production possible, and the transformative power of electricity
appeared in new devices such as the radio, telephone, television, home appliances and electric
lighting. Additionally, the development of the internal combustion engine enabled the
automobile and the airplane, and all the new industries related with these new technologies.
The third industrial revolution, usually called the computer or digital revolution, began in the
1960s, and was mainly based on the development of semiconductors, mainframe computing
(1960s), personal computing (1970s and ’80s) and, finally, the internet (1990s)2.
The importance of these industrial revolutions lies in the ways in which they changed the
structure of the economic and social systems and the working experience and lifestyle of billions
of people. All this has been possible because the new technologies that have brought about
these changes have had a ‘disruptive character’, i.e., they have not been just incremental
improvements in the already existing technologies, but completely new technologies that have
displaced an established technology and shaken up the industry by sweeping away the systems
or habits they have replaced3. Thus, for instance, in the last forty years the personal computer
has displaced the typewriter and has changed the way we work and communicate. The way we
communicate has also been transformed by the e-mail, which has displaced letter-writing and
disrupted the postal industry and, needless to say, by mobile phones which, by making it possible
to call people wherever they are and take and send photos almost instantly, have made phone
booths and point-and-shoot cameras obsolete. The cumulative impact of these three industrial
revolutions and of these disruptive technologies has been an incredible increase in wealth and
opportunities.
Today we are at the early stages of a fourth industrial revolution, which began at the turn
of this century, and which builds upon the third or ‘digital’ industrial revolution. However,
compared to previous revolutions, it differs in speed, scale, complexity, and transformative
power. It is still difficult to predict which of the multiple emerging technologies is going to
become fundamental in this new era, but Artificial Intelligence (AI) is going to be crucial and,
probably, the most impactful emerging technology in interconnection with some of the other
emerging technologies such as blockchain, robotics, driverless cars, and the Internet of Things
(IoT). Artificial intelligence (AI) is increasingly expected to disrupt the ordinary functioning of
society, from how we teach and learn to how we govern society or fight wars, from how we
interact with others to the products we manufacture or the services we provide. As we already
know now, the impact of this emerging technology will affect most areas of human activity4.
1 See SCHWAB (2016); SCHWAB & DAVIS (2018).
2 SCHWAB (2016), pp. 11 et seq . See also XU et al. (2018), pp. 90-95.
3 The term “disruptive technology” was coined by Clayton M. Christensen (1952-2020), professor of the Harvard Business School, in his
paper BOWER & CHRISTENSE N (1995), pp. 43-53. See also, CHRISTENSEN (1997), where he separates new technology into two
categories: sustaining and disruptive. Sustaining tec hnology relies on incremental improvements to an already established tec hnology.
Disruptive technology lacks refinement, often has performance problems because it is new, appeals to a limited audience and may not
yet have a proven practical application.
4 LIU et al. (2020). See also J INDAL & SINDHU (2022), where the y also describe the main technical aspects of AI.
Para continuar leyendo
Solicita tu prueba