Electronic agents are helpful and without risks for business as well as consumers
Undoubtedly, the influences of IT (‘Information Technology’) have already invaded every corner of our daily lives. Nowadays, it is unimaginable if one determines not to relevant with this new technology at all. We take cash from the automated teller machines (ATMs or cash point dispensers outside banks); we order train or coach tickets from their homepages; we pay with credit cards in store to avoid taking large amount of notes or coins. Not only the individuals, but the activities of organizations are also deeply affected. Electronic fund transfers (EFTs) transactions are made between financial institutions, including the employers paying salaries to our accounts; and a large manufacturing company may order components automatically and electronically from its suppliers when stock levels reach a predetermined lower limit. Traditional management is already out of date. It cannot compare with E-commerce, which is able to ‘maximize efficiency by reducing repetition and delays, increasing accuracy and permitting the maintenance of minimum stock levels by placing orders for ‘just-in-time’ delivery.’ As the U.S. ex-president Clinton said that, ‘[t]he invention of the stream engine two centuries ago and the harnessing of electricity ushered in an industrial revolution that fundamentally altered the way we work, brought the world’s people closer together in space and time, and brought us greater prosperity. Today, the invention of the integrated circuit and computer and the harnessing of light for communications have made possible the creation of the global Internet and an electronic revolution that will once again transform our lives.’
Among the countless achievements of Information Technology, agent technology is one of those most important inventions which invoked increasing attentions. ‘[I]t should be clear now that we live in an ‘agent’ world. All the time, where we are, we are surrounded by agents. Nobody is able to deal with all necessary tasks personally.’ The development of IT created a brand new type of agents: ‘electronic agents’. This sort of agents ‘can autonomously perform certain tasks in a network setting’. They ‘are autonomous, they are reactive and pro-active, they can adapt to their environment, they can learn and they may be creative.’ Although ‘[t]his does not imply that an electronic agent will possess all these characteristics, but it will possess some’.
The appearance of this new type of electronic agents is doubtlessly changing the existed legal opinions in agent law. Lots of arguments have been brought forward around this new business. Here is a sound that ‘electronic agents are helpful and without risks for business as well as consumers’. But it is still not clear that to what extent the electronic agents can take the place of human agents and what are the legal effects of the activities completed by electronic agents. Obviously the electronic agents are more efficient than human agents, but are they reliable? The task of this essay is to discuss the legal risks accompanied with electronic agents and to suggest the suitable reactions of law.
PART II Structure of this Article
To illustrate the legal problems caused by electronic agents, obviously, we should begin with the topic we are discussing, the definition of electronic agents and their principium. Therefore, the first part of the article will try to give out a definition via comparing three widely accepted definitions. Then, I will introduce the working principle of agent technology through analyzing how electronic agents work. Certainly, only when we make sure the basic principle of electronic agents, we can analyze the reliability and potential risks of this new business.
The following part aims to illustrate the possible risks of electronic agents and the challenges to our existed legal rules. Apparently, no one can deny that the emergence of electronic agents has brought us lots of benefits, such as less cost, more efficient, and so on. ‘The economic potential of the Internet is as difficult to predict at the beginning of the twenty-first century as the economic potential of the motor car was at the beginning of the twentieth century.’ But, generally, science and technology is a coin with two sides, for instance, the power of nuclear, which can be used for peaceful purposes and, on the other side, can be used for massacre. So the law should not hamper the development of this new technology, but at the same time it should encourage this new business and induce it towards a brilliant future via avoiding its potential risks.
As to the drawbacks of the electronic agents, the risks normally came from two origins: agents and agent platforms. Because of the limits of technology, the potential damages cannot be avoided in the agents themselves and also in the platform on which electronic agents base and communicate data and information with each other. Those risks include modifications of the task of agents, the security of confidential information, and the improper managements from agent platform, attacks to agent platform from malicious or faulty agents, and other numberless possible risks. All in all, ‘it became clear that the use of technologies like agents may cause legal problems’ and ‘[i]t is our belief that these legal issues are especially important when agents are developed to perform certain legal tasks or to support legal practice.’ Therefore, the opinion, which believes electronic agents are totally not risky for business as well as consumers, is obviously too optimistically. In fact, there are too many risky factors for this essay to discuss all the legal problems caused by them. In all of those legal problems, the most important one is the legal status of electronic agents. If the legal status of electronic agents can be accepted as same as the human agents, then the legal rules for human agents can be implied to electronic agents.
There are several totally different opinions about the legal status of electronic agents. I agree that although ‘the method of execution is case-specific’, the electronic agents have ‘no principal differences’ with human agents. We can see the electronic agent as an independent ‘electronic person’ and allocate the rights and duties between them, their principals and third parties as what we have done for the human agents.
Finally, the conclusion part will repeat the author’s attitude towards electronic agents. The technology of electronic agents is a skean, and there is no necessary to be optimistic blindly or to be pessimistic blindly as well. ‘Agent technology is progressing at a fairly steady pace neglecting reflections on legal aspects.’ Law should not, and also cannot, stop the steps of science and technology. What we can do is to predict the influences of new businesses, then try to limit those drawbacks and induce they develop healthily towards a correct direction.
PART III Electronic Agents and Their Working Principles
3.1 What is an Electronic Agent?
In fact, ‘electronic agents’ is not a strange thing for us. Even if we haven’t realized, we contact with electronic agents regularly. There is a wide variety of electronic agent types, ‘ranging from the more simplistic search engines, to personal assistants, to buyer and seller agents which respectively assist buyers and sellers in identifying and comparing the characteristics of goods and services in the electronic marketplace, to the more autonomous, intelligent agents.’ The most often seen sort of electronic agents is the search engines, such as google.com, openfind.com, wisenut.com and others. With the help of search engines, the users do not need to browse every website one by one. All they have to do is to input the keywords, and then search engines will present the result links which include the keywords users required. In this case, search engines represent the user to do the job of searching web-pages one by one. So, search engines serve as the users’ electronic agents.
However, although we are familiar with electronic agents, ‘[t]here is no generally accepted definition of an ‘electronic agent’.’ Not only no generally accepted definition, but also no a uniform term is use to refer this new business. Normally, they are known as ‘electronic agents’, but sometimes they are called ‘software agents’, and when compared with robots (hardware agents) , they are also named ‘intelligent agents’.
Though no generally accepted definition exists, it is unavoidable to mention three widely noticed definitions. The first one is the definition of the U.S Uniform Electronic Transaction Act (UETA). In section 2(6) of UETA (1999), it defined ‘‘Electronic agent’ means a computer program or an electronic or other automated means used independently to initiate an action or respond to electronic records or performances in whole or in part, without review or action by an individual.’ Another definition is given by the U.S. Uniform Computer Information Transaction Act (UCITA), which provides: ‘‘Electronic agent’ means a computer program, or electronic or other automated means, used independently to initiate an action, or to respond to electronic message or performances, on the person’s behalf without review or action by an individual at the time of the action or response to the message or performance.’ Except the regulations in U.S.A, in 1999, the Uniform Electronic Commerce Act (UECA) adopted in Canada gave its own definition: ‘‘electronic agent’ means a computer program or any electronic means used to initiate an action or to respond to electronic documents or actions in whole or in part without review by a natural person at the time of the response or action.’
According to those three widely known definitions, we can find there are no substantial differences. All of the three definitions declared ‘the electronic agent means a computer program or other electronic means’ which can be used to initiate an action or to respond to electronic records or actions and meanwhile review by human is not necessary for electronic agents. The differences are only that UETA did not require clearly no human interferences at the time of action or response but UCITA and UECA required, and UETA and UECA expressed explicitly that the automatism of electronic agents may ‘in whole or in part’ but UCITA did not mention that.
All in all, electronic agents are those electronic methods which can represent another person to complete certain tasks. The most important feature of electronic agents is automatism; therefore they can be separated with human management partly or totally. The extent of electronic agents’ automatism is determined by the technology. ‘[T]here is much ongoing research to develop autonomous, intelligent agents that would be capable of initiating action, negotiating and concluding contracts without the knowledge or oversight of the human being who deployed such agents. Furthermore, nowadays it is not only possible for agents to assist in the negotiation and conclusion of contract on behalf of a party, but also to perform part or, in some cases, even all of that party’s obligations in the contract.’ No mater to what extent the automatism of the electronic methods is, those programs with the feature belong to electronic agents. To sum up, ‘electronic agents’ refers to ‘programs that react autonomously to changes in their environment and solve their tasks without any intervention of the user.’
3.2 How Electronic Agents Works?
It is a problem for one without particular IT knowledge, such as me, to explain the working principle of electronic agents. In brief, electronic agents can be divided into two categories according to their different working methods: stationary agents and mobile agents. Meanwhile, both stationary agents and mobile ones have the similar working procedure, i.e. the program as an agent must be developed and then be transferred to a platform to achieve its tasks. So, agents and agent platforms are indispensable components in agent technology.
At first, we can divide electronic agents into two categories to analyze their different working methods: stationary agents and mobile agents. As their names express out that, ‘stationary agents are not able to leave their original environment, whereas mobile agents are software programs that move around (migrate) independently in heterogeneous computer networks.’ It’s not hard to understand the working method of stationary agents. In fact, the electronic calculator we regularly use is one of the examples of the simplest stationary agents. The task of the program installed in electronic calculator is to deal with the data input. This process stays in a limited platform, and calculates the data passively. Obviously, the advantage of stationary agents is the stability. Since that the work of stationary agents is limited into a platform, they are not easily affected by other environments. And even if they are attacked by hostile programs, such as the computer viruses, it is easy for the users to handle that. After all, the stationary agents are stationary and under the control of the user at any moment. But things are quite different with mobile agents. After being allocated tasks, mobile agents will migrate to other platforms to collect data and continue their jobs. Mobile agents ‘are autonomous and pro-active, can autonomously roam the Internet, perform transactions, and gather information.’ And ‘[m]obile code is executed in environments that may be completely unknown to the owner of the agent or where they possibly can have little influence.’ Some mobile agents even have the functions to clone themselves and allocate tasks to cloned agents, and then those new agents migrate to other environments else.
After they completed their tasks, the mobile agents then move back to the user’s platform with the data they have collected. ‘Therefore an infrastructure of agent servers is necessary, which can dispatch, receive, and implement these agents.’ Compared with stationary agents, the mobile ones are more efficient and, certainly, harder to be regulated. Mobile agents have a lot of incompatible advantages, which include ‘decreasing network traffic by performing computational intensive processing near the server that provides the source of information, easing transportability of services across providers, increasing the network availability by autonomy and asynchronous agent operations, reducing time and effort for installation, operation and management, enabling ‘on demand’ provision of special services, and allowing for a more decentralized realization of management and service control and therefore reduce the dependence on network availability resulting in a more robust system.’ But because of its migration, it’s also hard to manage the behaviors of mobile agents. After being injected into the network, mobile agents are nearly out of control from the owner. They will wander across from sever to server and eventually return to the user when their task has been completed. Because ‘the code and data of a mobile agent do not reside on the computer system of the agent’s owner, but on another computer’ , it is nearly impossible for the owner to influence its activities. If the mobile agents can clone themselves or work anonymously, things will be more complicated.
From another view, both the stationary electronic agents and mobile electronic agents cannot work without two essential bases: the electronic agents and the agent platform. ‘Agent technology refers to not only software agents, but also agent platforms: supportive middleware which enables the functioning of software agents.’ Or in other words, agents are ‘usually envisioned as ‘active entities’, while agent platforms constitute a technical (software) infrastructure to support agents. Their responsibilities differ: agents pursue goals for user applications, while agent platforms manage agents and their needs for resources.’ Electronic agents are autonomous and pro-active programmes, which can communicate with other agents, e.g. negotiate on prices and quantities for goods, and can interact with their environments, e.g. achieve the task of collecting data from the environments. Meanwhile, electronic agents can act only stay in an environment: the agent platform. For a stationary agent, the agent platform normally is the user’s computer system; while for a mobile agent, the agent platforms are the different systems which moved to. The agent platforms generally have some functions of managing agents, providing communication and interaction services and offering migration services, and so on. For instance, ‘[a]n agent platform manages the allocation of resources to agent, e.g. such as which agent ‘runs’ on which computer (if a choice exists) with what speed and which services (e.g. http-access to remote web-site, or access to a local database).’ And ‘[a]n agent platform provides mechanisms for security, including access control for migrating agent as well as defence against possible malicious, or faulty, agents.’ Also, an agent platform provides communication services, e.g. reliable and secure communication protocols only between particular agents.
Therefore, all the potential risks emerge out either from the stationary agents or mobile agents. Or from another perspective, all the risks are relevant either with the electronic agents themselves or with their agent platforms.
Potential Risks and Challenges to Legal Rules
4.1 Possible Risks of Electronic Agents
There is no necessary to list out the benefits electronic agent technology has brought to us. Just image the burdens which our ancestors have suffered when they tried to solve some complex mathematical problems. And how boring will it be if we must search a keyword via browsing every homepages by ourselves? Apparently, no body can deny that electronic agents are quite helpful. After all, there are lots of specific tasks which humans are either not able, nor willing, or even not authorized to implement on their own.
However, ‘[a]pplication of any technology has consequences, some of which are foreseen and wanted, others may be unwanted. Agent technology is not different in this respect: its application may involve certain risks.’ Compared with human agents, the possible risks of electronic agents are also relatively obvious. We can divide the risks into two categories based on the working principle of electronic agents: risks in agents and risks in agent platforms.
As to the agents, the risks may emerge out at two main periods: agent may be corrupted during agent storage and may be corrupted during transfer. During these periods the agents are stored or transferred, they may suffer possible attack and finally bring to an unwanted result. As we mentioned before, mobile agents suffer more potential risks than stationary agents. ‘Agents are expected to function ‘autonomously’, without a user’s constant supervision. Agents run on an agent platform, communicate with multiple agents on the Internet and interact with numerous objects and services are involved. Mobility increases the risks an agent faces: agents may be hosted by a potentially malicious agent platform, outside the control of their user and thus become more vulnerable…the other parties, such as agent platform, other software agents, humans, virus applications, etc., can change or copy the code and/or data of an agent, either intentionally or accidentally…..For example, an unwanted modification of an agent’s code and/or data may influence an agent’s behavior significantly. Changing data especially confidential information or data with an economic value may be sufficient to cause significant damage. In addition, instructions and goals of an agent can be modified.’ Apart from the risk of modification, ‘[a]nother issue is stealing information, including private information such as addresses, bank accounts, credit card numbers, and an agent’s internal (strategic) aims and instructions.’ More incredible thing is that ‘an agent may disappear temporarily or permanently, possibly representing a loss of investments in training and acquisition of the agent as well as frustrating the acquisition of results from the agent.’ In spite that the development of technology already makes electronic agents today more reliable, we still cannot say the agent technology is stead enough to worry about potential risks. At the same time the computer hikers’ technology are developing as well, and generally, hikers are smarter than well-educated computer engineers.
As to the agent platforms, they also face possible risks. ‘An agent platform hosts an agent: the code and data of which validity, reliability and trustworthiness cannot be easily determined automatically.’ As a result, ‘[c]ode and/or data of an agent platform can be modified or stolen, by mistake or by design. Possible consequences include modifying or stealing information on agents or conditions for (dis)allowing access to migrating agents, or management policies, etc.’ Not only the potential attack from hostile agents or viruses outside, the agent platform itself cannot promise no risks exists. Agent platform is a computer system in essence. The reliability and stability of platforms are also problems deserve to be considered. We can feel this easily just from the situation we often suffered when sending some e-mails. The servers for e-mails are also platform. And because of the instability of the servers, we usually find the letters to inform us ‘the delivery is failed’. To date, from the technological perspective, the potential risks are unavoidable in agent platforms.
Since the risks always exist, both in the agents themselves and in the agent platforms, the applications of electronic agents are not without risk. On the contrary, electronic agents may face more and new risks that human agents face. Therefore, a lot of brand new legal issues are caused.
4.2 The Suitable Attitude of Law
Regarding to the risks accompanied with electronic agents, numerous legal issues are caused. Such as the criminal obligations and civil obligations of illegal modification of electronic agent, the legal relationship between electronic agents’ users and providers, the protection for private information involved during the use of electronic agents, the patent rights of the electronic agents, the response of the damages caused by potential risks, and so on.
It is impossible for this essay to discuss all the legal issues above. As I see that, in all of those legal problems, the most important one is the legal status of electronic agents. Because the legal status of electronic agents is not clear, it is also not clear whether the existed agent-principal legal principles can be implied to electronic agents. And also it is a puzzle who should take responsible for the legal results caused by the behaviors of electronic agents. There are several totally different opinions about the legal status of electronic agents. One is called ‘traditional approach’, which regards the electronic agents as only tool or methods of users. The shortcoming of this opinion is very obviously. Since electronic agents are autonomous to some extent, they do not simply follow the determinations of users. After all, an agent should be a person having power to act on behalf of another and affect that other’s legal rights and liabilities.
The traditional approach equals electronic agents with tool; then the electronic agents are not ‘agent’ any more. The second opinion is called ‘modern approach’, which supports ascribing the electronic agent as a legal person. But, to date, ‘no legal system regards electronic agents as separate legal persons distinct from their users.’ It is still a question that can program be independent legal person? If we admit electronic agent’s legal personality, lots of problems will follow, e.g. when this legal person begins and what are the rights and obligations of this legal person. Another thought claim to regard electronic agent as a person with limited capacity, viz. they have contractual capacity but without legal capacity. This attitude is not reasonable with the development of agent technology. Nowadays, the functions of agent are not limited into the scope of negotiations or collecting information and data. Sometimes, both sides are electronic agents and they negotiate and perform the contract even with out interference of human users. In that case, the electronic agents are not methods with only contractual capacity. The fourth attitude is ‘the progressive approach’. This approach protests an independent ‘electronic person’, who is not a natural person and not legal person as well. There could be an agent register. ‘The owner of an agent could grant a certain amount of money to the agent by enrolling it into this register. The result would be a kind of agent with limited liability.’
I agree with the fourth approach. Since that the electronic agents are autonomous agents, they are not simple tool of users, to some extent. As an agent, an electronic agent can have its intention and, as a result, should take relative legal responsibilities. It’s not suitable for them to have legal personality at this moment, but we can let them have some responsible capacity via registration. Undoubted, the implement of electronic agents must face lots of potential risks. However, the requirements of registration granting a mount of money to register can reduce the risks efficiently. Even if the possible risks happened, the amount of money can be used to redeem the damages. All in all, although ‘the method of execution is case-specific’, the electronic agents have ‘no principal differences’ with human agents. Through registration, the electronic agents can be seen as independent ‘electronic person’ and we can allocate the rights and duties between them, their principals and third parties as what we have already done for the human agents.
Every step of science and technology is deeply changing our world. Similar as countless technology achievements, electronic agent technology also affects from two sides. Out of question, electronic agents are helpful. Meanwhile, ‘[t]he list of potential problems is almost innumerable.’ So, there is no necessary to be optimistic blindly or to be pessimistic blindly as well. What the law should do is not to limit the development of new business, but to reduce the possible risks and to encourage this new technology be utilized healthily.
1. David I Bainbridge, Introduction to Computer Law, 4th edn. (Person Education Limited, 2000)
2. William J. Clinton, The White House, Office of the Press Secretary, ‘Text of the President’s Message to Internet Users’, http://www.whitehouse.gov/WH/News/Commerce/message.html, 1st July1997
3. Artificial Intelligence and Law (2004) 12: 1-3, Foreword
4. Frances Brazier et al., Law-abiding and Integrity on the Internet: A case for Agents, Artificial Intelligence and Law (2004) 12: 5-37
5. Emily M. Weitzenbock, Good Faith and Fair Dealing in Contracts Formed and Performed by Electronic Agents, Artificial Intelligence and Law (2004) 12: 83-110
6. Steffen Wettig and Eberhard Zehendner, A Legal Analysis of Human and Electronic Agents Artificial Intelligence and Law (2004) 12: 111-135
7. Yaman Akdeniz, Clive Walker and David Wall, The Internet, Law and Society, (Person Education Limited, 2000)
8. Brenner, W., Zarnekow,R., and Wittig,H., Intelligent Software Agents: Foundations and Applications, (Springer, 1998)
9. The U.S Uniform Electronic Transaction Act (1999)
10. The U.S. Uniform Computer Information Transaction Act (1999)
11. The Uniform Electronic Commerce Act (1999)
12. Kotz, D. and Gray, R.S., Mobile Agents and the Future of the Internet, ACM Operating System Review, (1999) 33(3): 7-13
13. Larry Korba, Towards an Agent Middleware Framework for E-commerce, Netnomics 2 (2000): 171-189
14. Robert Bradgagte, Commercial Law, 3rd edn. (Reed Elsevier (UK) Ltd, 2000) p.128
The author can be reached at: SachinM@legalserviceindia.com