The Necessity of Building a Machine That Is Capable of Distinguishing Right and Wrong - Essay Example

The Necessity of Building a Machine That Is Capable of Distinguishing Right and Wrong In today’s world, computers are capable of approving financial transactions, driving cars and even controlling electrical supplies. At this rate, it is therefore anticipated that soon service computers will be located in the elderly homes where they will be capable of taking care of the elderly. For this reason, Colin Allen and Wendell Wallach argue that because of the anticipated increased in the responsibility of robots in our lives, they should be programmed with moral decision-making abilities. This increased capabilities will enhance their efficiency and our safety too. These authors acknowledge the necessity of building a kind of functional morality which will enable these artificial moral agents to possess some substantial ethical sensitivity. Further, these authors indicate that the quest to develop machines that can distinguish right from wrong has commenced. Robots and computers, therefore, can be programmed to have moral and ethical capacities. This is because some existing artificial intelligent systems exhibit to some levels this ability, for example, the autopilots and the framework based on Franklin’s LIDA which is capable of making moral decisions. But again, what is the necessity for building a machine that is capable of distinguishing right and wrong? Artificial moral agents (AMAs) does not only prove to be necessary but to some extents inevitable. The recent mushrooming of robots almost in every sector in our nation is a clear demonstration of the market pool for such innovations. Additionally, the freedom provided for their development talks no less, as Rosalind Picard, Vyzas, and Healey (1180) so aptly put it, “The greater the freedom of a machine, the more it will need moral standards.” This is because the increased innovative technologies of sophisticated systems such as those with moral capabilities require that they are fitted with abilities for moral decision making. Even though Colin Allen and Wendell Wallach acknowledge that at present, there are a few chances of an artificial moral agent that can be compared to human beings, there are future possibilities of such systems. This is because complex and more pervasive stems such as automated aircrafts and automated weaponry have demonstrated the transfer of human-decision making capabilities to machines. These systems like the automated aircraft, the authors explain, are developed to observe some levels of accuracy. For example, autopilots designed to observe some degree of turning radius can be regarded as having some moral functionalities. This is because these autopilots implement some ethical concern for passenger’s safety and comfort. To ensure this standard, most of the autopilots are designed to observe a specific turning degree that will not frighten the passengers or make them uncomfortable. To some level, this can be regarded as an AMA that is programmed to attempt observing some moral considerations. What measures can ensure the development of a fully functional artificial moral agent? In order to develop an artificial moral agent, first, there is the need bridge philosopher’s thought of abstraction and Engineers beliefs in the buildable designs. This is because theories are capable of informing designs just like various designs can formulate theories. The implementation of these moral capacitates in robots as indicated by Wendell Wallach lies within two broad approaches; top-down imposition on an ethical theory and bottom-up development of systems that can improve their own performances irrespective of these standards specified theoretically. The difference between these approaches is that top-down imposition on ethical theory method depends on already formulated rules. Example of such rules includes; Golden Rule and the deontology of Kant’s categorical imperative. These rules aid during the development and to later determine the performances of these AMAs. Contrary to this opinion bottom-up development method employs an evolutionary and developmental process in the development of AMAs (Colin Allen and Wendell Wallach, 58). If followed correctly these approaches can aid to the development of an efficient AMA, for example, the top-down approach enables a broad definition of ethical values. This extended scope will enable these AMAs to make a variety of ethical decisions and cover countless situations. Moreover, the bottom-up approach is capable of dynamically integrating inputs from distinct systems. This kind of automation will make these AMA reliable and more cost-effective. Colin Allen and Wendell Wallach identify that either top-down or bottom-up approaches can be useful in developing AMA capable of addressing moral challenges. But they emphasize the creation of a hybrid AMAs with improved capabilities. These Hybrid AMAs will employ both the top-down and bottom-up approaches in their development. In implementing both of the approaches, these AMA will be able to demonstrate flexibility and dynamicity provided by bottom-up approaches with diverse inputs capabilities while ensuring that decisions made are subjected to top-down principles. These abilities will enable hybrid AMAs to attain ideals that human being possesses. Additionally, in the attempt to develop these AMAs, Colin Allen and Wendell Wallach say that “Evaluations of whether the behavior of an artificial moral agent is acceptable will be made by applying criteria similar to those people use for evaluating each other’s behavior.” This statement is right since morality is human-centered. Therefore the early stages for developing artificial moral agents have been focusing on the simulation of human cognitive faculties. Simulating artificial moral agents requires a high degree of specificity of all human moral behaviors. In addition, specialists have discovered that simulation of complex programs especially those that involve human abilities requires a thorough analysis of every detail. This is because poor or incomplete simulations can result in AMAs’ failures or display of unanticipated results by AMAs when presented with new variables. The process of building an AMA require observation of some key mechanisms. As Wendell Wallach puts it “Building an AMA draws attention to the importance of cognitive mechanisms other than the capacity to reason for the selection of appropriate courses of action.” This mechanisms aids in various tasks in the development of an AMA. For instance, they activate influences on judgments and are also responsible for providing information that can be used in arriving at decisions. Understanding these mechanisms does not only help in the development of AMAs but also aid in understanding some technical concepts around AMAs. For example what triggers an agent to identify that it is in a position to make a moral judgment, how will the agent differentiate between more important information and a less important information that should be considered in making this judgements and how does an entity capable of making a moral judgment behave considering that these entities lack self-consciousness, emotions, and social skills. For these reasons, the development of an effective and more efficient AMAs will take a substantial amount of time. This is because a practical application must be developed one which exhibits decision making abilities in different contexts. Additionally this systems must be tested to realize opportunities and shortcomings arising from improper simulations of specific cognitive mechanisms. Michael Fisher, Christian List, Marija Slavkovik, and Alan Winfield on their study “Engineering moral Agents” indicates that Artificial Intelligent (AI) systems differ fundamentally from human beings because moral agents are entirely attributed to humans in the moral philosophy. In their argument, they elaborate that machine’s components differ from human biology and for this reason human beings possess characteristics that are lacking in AIs. For example mortality, self-consciousness, emotions, and the ability to learn from others. It is therefore impractical to transfer such knowledge especially those that are acquired through traditional lessons to the artificial agents. In addition to these concerns, preliminary investigation on the various models of moral reasoning suffers from issues such as; the frame problem which emanates from lack of psychological knowledge and the knowledge that enable identify consequences of some actions. Because of the frame problem, AI can experience difficulties in making moral decisions. Moreover, because the AMA will factor through a vast information of moral considerations, designing an AMA will require calculated interaction between several mechanisms. It is difficult to factor explicitly mechanisms utilized during decision making from those mechanisms that aid through the decision making process and those that continuously provide information currently being processed. These cumulatively make it impossible to develop an AMA in the view of Fisher, List, Slavkovik, and Winfield. In conclusion, this document herein has presented comprehensive ways of thinking towards developing a fully functioning AMA. A comprehensive model of an AMA will demonstrate the ability of a computer to make a moral decision. To this level, human abilities in making moral decisions will be represented computationally. As Wallach, Franklin, and Allen acknowledge the current absence of a comprehensive model for moral decision making, they remain optimistic about the existence of such systems in the future. In the attempt to explain their position Wallach, Franklin, and Allen suggest a framework based on Franklin’s LIDA (Franklin and Patterson 2006). Further, the capabilities exhibited by autopilots can be a clear example that robots and computers can be imparted with human knowledge especially those for making moral decisions. The Franklin’s LIDA model will be capable of accommodating various moral considerations and demonstrate how these considerations can help in making moral decisions collectively. However, even though this proposed model is just a rudimentary there are chances of developing a much more comprehensive framework in the future. With the increasing advancements in artificial intelligent systems, moral conduct of machines is a growing concern especially to the societies experiencing these advancements? It is therefore important for all the stakeholders to involve in the development of these programs for example engineers, philosophers, psychologists, and other experts to collaborate in the implementation of reasoning in these machines. Designing moral machines should not be entrusted to the manufacturers alone but also should be subjected to public scrutiny through initiatives such as seminars and other public discussions on the challenges of machine ethics. These initiatives are important in ensuring that this AMAs become reliable and gain public trust. Additional research directed at implementation of AMAs will not only contribute to the ongoing discourse but also will also suggest other comprehensive models capable of making a moral decision. Read More