Will and declaration in acts performed by intelligent software agents first notes on the question Francisco Andrade 1, Paulo Novais 2 and José Neves 2 1 Universidade do Minho, Escola de Direito, Braga, Portugal fandrade@direito.uminho.pt 2 Universidade do Minho, Departamento de Informática, Braga, Portugal {pjon,jneves}@di.uminho.pt Recent trends in the field of Artificial Intelligence, brought along new ways of contracting, through the use of electronic agents and what we could call intelligent inter-systemic electronic contracting [1][2], or situations of contracting through the only intervention and interaction of autonomous informatics intelligent devices, capable of acting, learning, modifying instructions and making decisions. [3] An agent is a software program that acts flexibly, on behalf of its owner, user or client, to achieve particular objectives. To do this, they must exhibit the following properties [4, 6]: autonomous: capable of making decisions about what actions to take without constantly referring back to its user; reactive: able to respond appropriately to the prevailing circumstances in dynamic and unpredictable environments; proactive: able to act in anticipation of future goals so that its owner s objectives are met. By increasing the level of the automation, use and roles played by the agent in e-commerce environments, the commerce becomes more dynamic, personalized, and context sensitive. Past Today The Future Assistant Mediators Initiators Figure 1: Agent roles Agents (initiators), today and even more in the future, will be active participants in the trading process: initiating and generating agreements without Andrade F., Novais P., Neves J. Will and declaration in acts performed by intelligent software agents preliminary issues on the question, Proceedings of the LEA 2005 - The Law of Electronic Agents, Oskamp A., and Cevenini C. (eds), IAAIL series, Wolf Legal Publishers, ISBN 90-5850-505-7, pp 53-56, 2005.
the individual being aware that contract negotiation has been initiated, let alone concluded (Figure 1) [5]. The intervention of such intelligent devices in electronic commerce brings along a radical shift in the way we understand basic legal questions such as will and declaration. Declarations of will and agreements will therefore no longer be generated through machines but by them, without any intervention or supervision of an individual [7] and, in this context, it will be interesting to analyse beyond the question of the expression of consent in itself, the issue of the declaration and of an eventual non correspondence between the will and the declaration. And a crucial question will be, in case of autonomous intelligent devices, how to deal with the divergences of will and declaration and whether or not it is possible to talk about defects of the will in declarations enacted by electronic agents. [8] The declaration of will is constituted by two different elements the external element (the declaration itself) and the internal element (the will itself, the real source of the declaration), and the two usually are coincident [9]. However, in certain situations there is no coincidence between what was really intended and what was expressed, between the real will and the declared [10]. On the other hand there is also a possibility that there was a coincidence between what was wanted and what was declared, but the will was indeed conditioned by defective motivation, considered by law as an illegitimate motivation of the will [8][9] This may well also happen with software agents declarations, but in this case some authors state that the will is not in the electronic agent, but in the person or entity in whose interest the agent is working for [11]. Yet, even considering the notion of attribution, proposed by Weitzenboeck in the sense that the acts of the agent must be attributed to the human using it [7] -- the truth is that no one would be in condition of anticipating the contractual behaviour of the agent [12], or even the good faith of the agent s acting considering the fact that the user may not have been directly involved or consulted by the electronic agent in the performance of the contract [13]. The Portuguese law on electronic commerce (Dec. Lei 7/2004 from the 7 th of January 2004) refers in art. 33 the electronic contracting without human
intervention (or inter-systemic electronic contracting as we call it), and it establishes that the common norms on error will be applied only in certain situations: If there is a programming error, it will be applied the norms on error on the formation of the will (art. 33, nr. 2 a); If there is a defect in the functioning of the machine, it will be applied the norms on error in declaration (art. 33, nr. 2 b) ; If the message does not arrive to the destination exactly the way it was sent, it will be applied the norms on error on transmission (art. 33, nr. 2 c). So, according to the Portuguese law it is possible to say that these are the recognised situations that lead to the application of the rules on error to electronic inter-systemic contracting. But, it is well known that the divergences between the will and the declaration may be understood in two different ways: as intentional divergences (voluntarily assumed by the author of the declaration) and as non intentional (and thus non voluntary) divergences [9]. And the way the Portuguese law on electronic commerce was written, it is undoubted that it was assumed that only non-intentional divergences will be considered in intersystemic electronic contracting, and even these in a quite restricted way. Considering this, the ultimate question must be asked: should the legal systems consider the daring possibility of an electronic agent (even if considered as a mere tool or instrument someone is using) having a will of its own? In fact, we are facing new forms of autonomous intelligent behaviour through the use of complex intelligent devices, capable of operating all alone and by themselves without any human intervention. Computer systems are now emerging that can operate not just automatically but autonomously [3]. And the characteristics of the newest software agents are becoming so sophisticated, that we must already face the possibility of software agents expressing emotions, or manifesting certain features of true personality [14]. And the question is how far can we go in considering computer intelligence and autonomy, how can we legally deal with the intervention in our daily lives of a new form of electronic behaviour such as the one of an intelligent and autonomous electronic agent or computer system situated in some
environment, and that is capable of autonomous action in this environment in order to meet its design objectives? [5] Considering this, we must recognize that intelligent electronic agents indeed have a capacity of learning, of reasoning, of choosing, of deciding and even of taking an initiative. And even when they act on behalf of a human person, even if they could be seen as a nuntius [15] [16], they may well go beyond that, as they have the capabilities to not just limit themselves to transmit a declaration issued by a person, but rather to have an autonomous reasoning and action. This is the departure point for a further study, intended to analyze whether or not it is possible to foresee the application with the required adaptations of the general law rules on divergences between will and declaration, and also of the general rules on defects of the will, to the declarative processes conducing to contractual declarations enacted by intelligent software agents. References: [1] -- Erica Brandini Barbagalo, Contratos eletrônicos, Editora Saraiva, São Paulo, 2001 [2] -- F. Andrade / J. Neves : Intelligent Electronic Inter-systemic Contracting Issues on Consent and Contract Formation in ICEIS 2004 Sixth International Conference on Enterprise Information System Proceedings, vol. 4, págs. 403-410 [ 3] -- Allen, Tom and Widdison, Robin Can Computer Make Contracts?, in Harvard Journal of Law and Technologie, Volume 9, Number 1, Winter 1996. [ 4] -- Andrade F., Novais P., Machado J., Abelha A., Neves J.: Legal Security and Credibility in Agent Based Virtual Enterprises, in PRO-VE'05-6th IFIP Working Conference on Virtual Enterprises, 2005 (to appear). [ 5] -- Wooldrige M., An Introduction to MultiAgent Systems, John Wiley & Sons, ISBN 0 47149691X, 2002. [ 6] -- N. R. Jennings and M. J. Wooldridge: "Software Agents" IEE Review, January, 1996, 17-20. [ 7] -- Weitzenboeck, E.: Electronic agents and the formation of contracts. In International Journal of Law and Information Technology, Vol. 9 Issue 3, Autumn 2001, Oxford University Press. [ 8] -- Mota Pinto, C.A. Teoria geral do Direito Civil, Coimbra Editora, 1976 [ 9] -- Andrade MD. Teoria Geral da Relação Jurídica, Coimbra Editora, vol. II., 1974.
[10] -- Heinrich Ewald Hörster, Teoria Geral do Direito Civil, A parte geral do Código Civil Português Teoria Geral do Direito Civil, Livraria Almedina, 1992, page 532 [11] -- Victor Castro Rosa, Contratação Electrónica, in Comércio Electrónico o novo regime legal, Gabinete Política Legislativa e Planeamento, Torre do Tombo, Lisboa, 22-4-2004, frame 16. [12] -- Giovanni Sartor, Agents in Cyberlaw, Workshop on the Law and Electronic Agents (LEA 2002). [13] -- Emily Weitzenboeck Electronic Agents and Contract Performance: Good Faith and Fair Dealing, Workshop on the Law and Electronic Agents (LEA 2002). [14] -- Ball, Gene / Breese, Jack : Emotion and Personality in a Conversational Agent, in Embodied Conversational Agents, edited by Justine Cassel, Joseph Sullivan, Scott Prevost and Elizabeth Churchill, The MIT Press Cambridge Massachussets, London England, 2000 [15] -- Hörster, E. E.: Sobre a formação do contrato segundo os arts. 217º e 218º, 224º a 226º e 228º a 235º do Código Civil, in Revista de Direito e Economia, Ano IX, Nºs 1-2 (1983) [16] Andrade F., Novais P., Neves J. Issues on Intelligent Electronic Agents and Legal Relations, in The Law of Electronic Agents Proceedings of the LEA 2004 workshop, edited by Claudia Cevenini Gedit edizioni, 2004, págs. 81-94