A Responsibility Ethic for Audit Expert Systems

Jesse F. Dillard Kristi Yuthas

ABSTRACT. To effectively pursue ethical action, the business community must recognize that the fundamental form of human association is not the social contract into which persons enter as atomic individuals, making partial commitments to each other for the purpose of gaining limited common ends or of satisfying certain laws. The fundamental form of human association is rather the face to face community in which ongoing commitments are the rule and in which aspects of every individuals experience are conditioned by the continuing membership. The following discussion initiates a preliminary phase in the consideration of what constitutes ethical issues associated with the business applications of expert systems. The focus is on knowledge based expert system applications in public accounting, specifically in the audit domain. Prior research on the development and use of expert systems in auditing has focused on a limited set of ethical issues. Niebuhrs theory of the the responsible self is used here to broaden the scope of what constitutes an ethical issue and provides a framework for identifying responsible action. Within this framework, an action is responsible if it takes into consideration ongoing relationships among the stakeholder groups affected. Actions prior to the development of the system along with Jesse F. Dillard is currently the KPMG Peat Marwick Professor of Accounting and Accounting Department Chair in the Robert O. Anderson School of Management, University of New Mexico. Professor Dillard received his Ph.D. from the University of South Carolina, has published widely in the accounting and business literature, and is currently working in the area of information technology and its ethical implications in the workplace. Kristi Yuthas is an associate professor at Portland State University (Portland State University). Her research focuses on the social and organizational impacts of information technology and management control systems.

the potential consequences for the system must be considered. The discursive requirements that provide the context and conditions necessary for implementing the proposed theoretical framework are presented and an illustration of how the responsibility ethic can be implemented in the audit expert system domain is developed. KEY WORDS: expert systems, information technology, responsibility ethics, stakeholder committee, systems development

Expert systems (ES) are computer systems designed to make expert level decisions within complex domains. The business applications of this advanced information technology has been varied and broad reaching, directed toward making operational, management and strategic decisions. Audit expert systems (AES) are such systems applied in the auditing environment within the public accounting domain. Major public accounting firms have been quite active in developing such systems (Baldwin-Morgan, 1993; Tomas, 1998; Yang and Vasarhelyi, 1998), and some argue that these tools and technologies will be increasingly important for survival as the firms strive to enhance their competitive position and to reduce their legal and business risk (BaldwinMorgan, 1998). The implementation and use of these powerful systems raise a variety of significant ethical questions. As public accounting firms continue to devote substantial resources to the development of AES, dealing with the ethical risks and potential consequences to stakeholders takes on increasing significance. Although the ethical implications of AES are beginning to be recognized, much remains to be done. As an

Journal of Business Ethics 30: 337359, 2001. 2001 Kluwer Academic Publishers. Printed in the Netherlands.

338

Jesse F. Dillard and Kristi Yuthas literature on AES ethics and Niebuhrs responsibility ethic (1963) including the conditions that must obtain for it to be most successfully applied. The second section discusses how Niebuhrs framework differs from current perspectives on ethical decision making in auditing and describes the process through which decisions are made under the responsibility approach. The third section provides an example of issues that might arise when the responsibility framework is used to analyze the ethical implications of an AES implementation.

initial step, this paper proposes an ethical framework for recognizing, discussing and evaluating AES implications for public accounting and its stakeholders. The paper is an attempt to develop and expand the ethical implications associated with the application of expert systems in business domains. The focus is on audit expert systems because the domain is relatively well developed and can be fairly easily delineated (e.g., see Vasarhelyi and Kogan, 1998). However, the implications of the discussion are germane for other large public service organizations as well as business applications in general.1 Previous research on ES ethics has been sparse and narrowly focused. Khalil (1993) considers the ethics of expert system applications in terms of the systems capacity for and legitimacy in moral decision making. He suggests that the legal or ethical responsibility for decision-making cannot be abdicated by humans (or organizations) through the implementation of expert systems, and he concludes that ethics cannot be transferred to systems; thus, the concept of artificial ethics is still science fiction (p. 319). Discussion within the audit domain has centered around the expert/firm relationship and tends to focus on legal implications (Specht et al., 1991) or ownership rights (Sutton et al., 1995) associated with audit expert system use. Following Dillard and Yuthas (1997), this paper expands the scope of these discussions by applying Niebuhrs (1963) theory of responsibility ethics. The normative arguments of the responsibility ethic allow for explicit consideration of multiple stakeholders within the audit domain. Application of Niebuhrs framework also provides a means through which ethical issues can be formulated with greater breadth and specificity. The purpose of this paper is therefore to show how a responsibility-oriented perspective can be used by audit firms and other public accounting constituents to identify a broad range of potential ethical issues and to address them in a systematic manner that considers stakeholder interests. As such, the responsibility ethic provides a broader basis for addressing ES applications within the audit domain. The remainder of the paper is divided into three sections. The first section describes extant

Theoretical background This section lays the theoretical groundwork for the responsibility approach to ethical dilemmas proposed in this paper. The section begins by defining expert systems and exploring past research on ethical issues associated with the use of ES in auditing. Previous research relies on a narrow interpretation of the theory of distributive justice, an ethical perspective that deals with questions surrounding the ethical distribution of rights and resources, such as audit expertise and its associated rewards. We argue that this perspective is limited and propose the use of an alternative approach for addressing AES dilemmas Niebuhrs responsibility-based ethics. This perspective is appropriate for AES decisions, because it is designed to address communal problems such as those associated with shared responsibility for system outcomes. Although Niebuhr provides a framework for analyzing the decision context to determine appropriate action, he does not discuss how an understanding of this context is to be gained by the decision maker. The final portion of this section describes the necessary conditions for a dialogue that allows stakeholders a voice in AES decisions.

Expert systems Expert systems and decision support systems can be seen as extremes on a continuum representing the extent to which a system possesses reasoning capabilities (Dillard and Bricker, 1992). A

A Responsibility Ethic for Audit Expert Systems decision support system contains no reasoning or logic capabilities but assists the decision maker by selecting information and sequencing decision processes. Expert domain knowledge is required on the part of the decision maker to arrive at the decision or judgment. In contrast, expert systems are designed to be used by decision makers who do not possess expertise in the problem domain. The human experts representation of the task domain provides the template for expert system design.2 The knowledge base and heuristic rules which are used to systematically search a problem space reflect the decision processes of the expert. A viable expert system is expected to perform this search as effectively and efficiently as a human expert. An expert system incorporates the reasoning capabilities of a domain expert and applies them in arriving at a decision. The system user needs little domain specific knowledge in order for a decision or judgment to be made. The users main decision is whether to accept the systems result. Decisions or judgments made by an expert system can be an intermediate component in a larger decision context. For example, an AES may provide a judgment as to the adequacy of loan loss reserves that an auditor would use as input for making an audit opinion decision. The fact that the output supports or provides input for another decision does not make the system any less an expert system. The distinguishing feature of an expert system lies in its ability to arrive at a nonalgorithmic solution using processes consistent with those of a domain expert.

339

The audit domain Baldwin-Morgan (1993), citing Brown (1991), states that currently, over 30 auditing expert systems are being developed and used by Big 6 firms at a cost of millions of dollars each year (p. 17). Using telephone interviews with Big 6 representatives, Brown and Murphy (1990) found that although there was a great deal of diversity among them, all of the firms were either using expert systems in their audit practice or had such systems under development. The respondents reported that expert systems were useful for

solving auditing and accounting problems. The functional areas in which expert systems were being used included audit work program development, internal control evaluation and risk analysis, and technical assistance. Audit-related systems or prototypes that have been developed include Expertest, ExperTAX, and RISK ADVISOR (Coopers and Lybrand, e.g., Graham et al., 1991), LOAN PROBE (KPMG Peat Marwick, Willingham and Ribar, 1988), CONTROL PLAN (Deloitte and Touche), RIC Checklist (Price Waterhouse), EDP-EXPERT (Hansen and Messier 1986), AUDITORS ASSISTANT (Srivastava et al., 1990), AUDITOR (Dungan, 1983), ICE (internal control evaluation, Kelly, 1984), XINFO (going concern opinion decision, Dillard and Mutchler 1988), TICOM-IV (Bailey et al., 1988) and AUDITPLANNER (Steinbart, 1984). While a critical review of extant AES is beyond the scope of this discussion, such reviews have been undertaken by OLeary and Watkins (1989), Brown and Murphy (1990), Meservy et al. (1992), Baldwin-Morgan (1993), Khani and Zarolwin (1995) and Tomas (1998). There is also a growing body of behavioral systems research that focuses on the design and use of expert systems in accounting (e.g., Steinbart and Accola, 1994; Pei et al., 1994; Eining and Dorr, 1991; Murphy, 1990; Gal and Steinbart, 1992; Pei and Reneau, 1990; Steinbart, 1987). Despite the increasing level of interest and investment in expert systems by public accounting firms (OLeary and Watkins, 1989; Baldwin-Morgan, 1993, 1998), little attention is paid to exploring the impact these systems will have on a firm and its stakeholders. This is not surprising, considering the dearth of literature investigating ethical problems relating to other types of information systems (Mitroff and Mason, 1989). Expert systems research both inside and outside the audit domain focuses largely on describing individual systems, exploring the benefits to be gained from expert systems and addressing technical complexities of developing successful systems. The primary justification for expert system development and implementation in general takes a decidedly utilitarian perspective. The primary ethical consideration is

340

Jesse F. Dillard and Kristi Yuthas authors frame the ethical issues in terms of distributive justice and consider the issues in light of three ethical reasoning categories: teleological, deontological and contractarian. They rightly argue that each of these ethical foundations provides a different perspective on the ethical dilemma being faced by an individual, a firm and/or society with respect to the development and implementation of expert systems in audit firms. As a result of their analysis, Sutton et al. (1995) propose the use of a contractarian framework to evaluate expert systems development and application. The arguments in prior AES research can be framed as problems of distributive justice. The central focus of theories of distributive justice is the question of how goods and opportunities should be equitably distributed. A broad array of answers to this question have been proposed. The perspective of AES researchers tends to coincide with Nozicks (1974) argument that rights to what one acquires are derived from how one acquired it. If the property has been legitimately acquired, its possessor has earned it, has the right to continue to possess it, and is entitled to any returns generated. The question of who is entitled to the use of expert knowledge thus hinges on whether the expertise was legitimately acquired by a person without infringing on the efforts or rights of others when the expertise did not previously exist, or whether the expertise was transferred from an expert who initially and legitimately acquired the expertise. If the issue is that of who is responsible for creating, or earning, the expertise, then under Nozicks proposal, the distributions of the rewards from its exploitation would be fairly easy to settle. Such is the position generally taken by Sutton et al. (1995) in all three of the positions they address.3 The problem with this perspective, and an important shortcoming of prior AES literature, centers on its view of legitimate entitlements. Any theory of distributive justice hinges on the question of who deserves what. As MacIntyre (1984) points out, Nozicks formulation presumes that one is deserving because one has earned the right. Other more liberal formulations (e.g., Rawls, 1971) presume that one is deserving because one has been deprived. Neither position

economic efficiency does it pay in terms of return on investment? The return can be obtained from reduced operating costs, increased efficiency, increased productivity, increased revenues and/or reduced legal liability risk. Apparently, experts, firms and researchers do not consider the implementation of expert systems as ethically problematic. The question of whether expert systems should be developed or implemented in particular circumstances is generally a non sequitur relative to the technological question of whether they could be developed. A limited number of recent studies has considered ethical issues associated with the development and application of audit expert systems. This work has taken either a legal perspective or an ownership perspective, and although it has advanced the discussion of systems development and use, the analysis has tended to be narrowly focused. Specht et al. (1991) raise the issue of who is liable when a system fails and conclude that participation in the development and use of an expert system can expose an auditor/expert to increased legal risk. The expert may be liable for audit failures attributable to the expert system if the knowledge base is found to contain errors or if the system is implicated in erroneous audit judgments. Several studies focus specifically on ownership rights. Ownership refers to rights or entitlements to specialized knowledge or expertise. Mitroff and Mason (1989) suggest that the development of expert systems will cause professionals to address the question of who owns an experts knowledge. Before the development of expert systems, expert knowledge leading to professional judgments resided in the minds of experts/ auditors and moved with them from job to job and firm to firm. When knowledge is incorporated into an expert system, an individual may lose ownership to, and control over, the expertise. Sutton and Byington (1993) identify the question of ownership as the primary ethical issue in expert systems applications. Sutton et al. (1995) focus on ownership as it relates to the replication and dissemination of an experts knowledge using computer software. These

A Responsibility Ethic for Audit Expert Systems considers the context of the distribution decision in terms of the ongoing relationships among individuals affected by the decision as does the responsibility ethic discussed in the following section.

341

Responsibility ethic When considering relations among ethical agents,4 one does not consider only the teleological ends being consciously pursued or the deontological self-legislated laws (e.g., social contracts) being obeyed. One considers how agents are responding to each others behavior based on interpretations of past actions and anticipation of responses to proposed actions, all grounded within a communal history. One interprets the things that are encountered as parts of wholes, as related to and as symbolic of larger meanings. (Murdock, 1992). Niebuhr (1963) argues that ethical systems must be formulated for, and by, members of an ongoing community. This is consistent with the perspective of normative stakeholder theory, which suggests that the ongoing rights and interests of community members should be considered before a firm takes action (Donaldson and Preston, 1995). Niebuhr proposes a responsibility-oriented ethic containing the following key elements: interpreted action, response, accountability, and social solidarity. Moral or ethical behavior requires awareness of the relatedness of events and can be viewed as a response to previous, interpreted action. When determining the appropriate response, the agent recognizes the condition of being held accountable for the consequences of that decision. These actions, responses, and consequences all take place within an ongoing community. Responsibility is derived from an agents recognition of his/her accountability for behavior and includes accepting the relational consequences as well as anticipating the continued reaction of the community. Accepting the continuing dialogue among agents forming a community is what Niebuhr refers to as social solidarity. Thus, responsibility based on personal accountability is dependent upon a relatively stable ongoing community of agents.

Niebuhrs perspective on the central question in ethics what shall I do? is cast in terms of an ongoing conversation among the agent and all affected parties. The responsibility ethic recognizes the involvement and connectedness of behavior and response. An agent responds to behaviors based on the interpretation of action as well as the anticipated interpretation of the responses of the other actors. The agent is held accountable for the response by the community of agents interpreting his/her act. The agent anticipates and accepts the reaction of the community of actors to the response. Niebuhr integrates a systematic dynamic into ethical considerations by recognizing the interconnectedness of actions, response and interpretation as well as the ongoing nature of the process. An ethical decision is not a one-time, isolated event. An agent is engaged in an ongoing conversation of behavior and response carried out within a sustaining community or society. The ethic of responsibility does not seek the good in a utilitarian sense or the right in a deontological sense, although it incorporates both of these ideals. The good and the right are defined within the process of determining which action is fitting, based on the history of the community.

Conditions for responsible behavior Niebuhrs responsibility ethic presents an abstract process-oriented perspective whereby ethical behavior is derived from an agent recognizing and being held accountable for his/her actions as part of a society. Niebuhr uses a descriptive metaphor of an ongoing conversation between the agent and the members of the community.5 The conversation metaphor has a literal component as well; implicit in Niebuhrs perspective is a reliance on the ability of community members to engage in open and trustworthy discourse. For Niebuhr, ethical behavior is predicated on an agents ability to communicate (i.e., interact and develop symbolic meaning) while participating as a nontransient member of an ongoing community. Thus, one of the first steps toward the implementation of a responsibility ethic is the stipulation of what constitutes legitimate

342

Jesse F. Dillard and Kristi Yuthas between an audit firm and its client with respect to the use of AES. The first validity claim requires that a behavior is the most effective and efficient means of obtaining an end. This claim is teleological in nature. The focus is toward instrumental, goal-directed behaviors. In using an expert system on an audit engagement, the audit firm claims, and the client legitimately accepts, that the system is technically sound, is appropriately applied, and represents effective and efficient service. The second validity claim is that a behavior is correct and proper in accordance with relevant norms. This claim is deontological in nature. The focus is toward behavior conforming to some covering law or set of predetermined standards. The audit firm claims, and the client legitimately accepts, that the system conforms to extant accounting and auditing standards and that the implementation of the system is within the scope of legal and/or professional guidelines. The third validity claim refers to the authenticity and sincerity of subjective behavior. The focus is toward behavior that is legitimate, in that the actual purpose of the agent is the stated purpose of the agent. This represents the critical assumption underlying the responsibility ethic of Niebuhr. An agent must be trusted to be genuine in his or her behavior. This claim moves beyond the teleological and deontological considerations. If an agent is not genuine in his or her assertions, strategic intentions are presumed. One agent is attempting to manipulate another agent for the formers gain. The trustworthiness of the dialogue is put in doubt. If any of these validity claims fails, legitimate, trustworthy dialogue can break down. Mutual understanding grounded on previous experience cannot be relied upon as the basis for responsible behavior. Intersubjective understanding is illusive and it may become difficult to integrate actions/responses into a larger context. Loyalty among the agents toward communal institutions may not develop. To pursue the engagement of a responsibility ethic within an audit context, it is thus necessary to approximate the requisite conditions for legitimate communication in which the validity claims can be made. To arrive at the desired goal

communal dialogue. According to Niebuhr, discourse requires an agent to assess the trustworthiness of the system of actors within which the agent is situated. Through ongoing discourse and action, the agent develops a sense of loyalty toward the community. Niebuhrs arguments suggest that if the communal discourse is distorted or controlled by powerful agents set on accomplishing self-interested goals, the ethic of responsibility becomes an ethic of manipulation and exploitation. Although Niebuhr recognizes the importance of fair discourse, the abstractness of his formulation makes it difficult to move from theoretical discussion to the actual implementation of the responsibility ethic in a practical domain such as audit expert systems. However, the responsibility ethic has similarities to contemporary stakeholder approaches and, like those approaches, can be useful in developing a broader perspective on the domain of interest. Normative stakeholder theory6 suggests that a firms stakeholders those groups potentially affecting or affected by a firms actions should have the opportunity to express their interests. The rights and responsibilities of these groups are to be considered before taking action if the action is to be an ethical one.7 To explore the conditions necessary to achieve fair discourse among stakeholders, we draw upon Habermas theory of communicative action. Habermas (1984, esp. 92104) stipulates that legitimate communication which forms the basis for ethical action requires the satisfaction of three validity claims: 1. propositional validity concerning external or objective characteristics (truth obliged to provide grounds); 2. normative validity concerning rightness relative to social norms (rightness obliged to provide justification); and 3. subjective authenticity relative to the perception and actual intention of behavior (truthfulness obliged to prove trustworthiness). These claims can be used to define the conditions under which the responsibility ethic can be legitimately pursued. In discussing these claims, we use as an example the interaction

A Responsibility Ethic for Audit Expert Systems of a dynamic, evolving ethic, certain conditions must be met. Following Habermas, Kettner (1993) identifies five conditions or characteristics required for an ethically rational dialogue: generality, autonomous evaluation, role-taking, power-neutrality, and transparency. Generality requires that a discussion must be open to all interested parties (or their representatives) who are competent to participate. Within an expert system context, everyone affected by the implementation and use of the system should have an opportunity to express an opinion. Groups affected by audit expert systems that should be included in the dialogue consist of, but are not limited to, experts, firm employees/users, firm management, clients, current and potential investors in clients, regulators, the accounting profession, and society. Autonomous evaluation implies that individuals should be allowed to interject their own individual interests and needs into a discussion and should not be restricted to seeking universalized ends such as the greatest good or equitable distribution. These individual interests are also open to criticism by other participants when, for example, they appear to be irrational or dogmatic. This implies that each of the stakeholder groups mentioned above is to be permitted to present its position with respect to expert systems applications as each would be affected, without fear of ridicule or reprisal but in anticipation of legitimate criticism. Role-taking requires that even as participants present their individual interests and needs, they transcend these self-interested positions so that they can understand the views of the other participants. For example, those responsible for the decision to develop an expert system would take seriously the interests of those affected by the system, viewing the circumstances from the perspectives of other stakeholders, such as the field auditor who will use the system or the client who must rely on the output. Power-neutrality suggests that participants must hold, or acquiesce to, positions of equal power with respect to participation and expression. Nondistorted communication can occur only when participants are not affected by asymmetric power arrangements. In an actual

343

dialogue between an expert and a firm, each may withhold certain statements for fear that negative consequences would result. Thus, to achieve undistorted communication it is necessary to develop mechanisms that allow parties to participate without fear of retribution; for example, the dialogue might take place under conditions of guaranteed anonymity or between experts of one firm and managers of another. Transparency reinforces the premise that most discourse relates to strategic behavior. Agents use language strategically to accomplish their own, self-interested goals. Therefore, for fair discourse parties must be encouraged to expose their own individual interests, strategies and goals and to seek to explore those of other participants. For example, if it is a firms intention to use an expert system to standardize audit practices and thereby reduce the work force, this purpose should be made known to all participants. Likewise, an expert seeking to obtain status within a firm would be required to reveal such objectives. If the goals are misrepresented or hidden from others, consensus based on full mutual understanding cannot be attained. Clearly, these conditions are highly restrictive and difficult to attain. Nonetheless, they provide ideals which can be pursued by firms or stakeholders seeking to identify moral courses of action when faced with ethical dilemmas. To the extent that the conditions for ethical dialogue are not met, the ethical consensus obtained from the communal discourse is brought into question and the responsibility ethic is incomplete or distorted. Thus, these conditions represent limiting conditions of the responsibility ethic. Firms committed to understanding and pursuing the ethical use of expert systems can use the constraints as guidelines for actual conversations. They can develop mechanisms through which stakeholders can be identified and invited to express their views in an ongoing and open discussion. The responsibility ethic formulated above does not specify universal weighting mechanisms for ethical outcomes. Each situation in which the approach is used would contain unique contextual elements. Thus, prioritization and selection of interests and outcomes could only be determined through legitimate dialogue among

344

Jesse F. Dillard and Kristi Yuthas facing a particular ethical dilemma. We reformulate this model and combine it with Niebuhrs ideas to develop a decision making framework designed to help AES proponents become aware of their stakeholders and to conduct ongoing, structured discussions with these stakeholders to identify potential ethical issues. Our analysis centers around the Lampe and Finn model because their model directly addresses ethical decision making in the audit domain. However, this model is based upon, and is similar to other general cognitive models, such as the one developed by Rest (1986). Lampe and Finn (1992) argue that the standard audit decision model, as set forth in the AICPA code of ethics, is not sufficiently complex to describe fully the processes by which auditors make ethical decisions. Lampe and Finn develop a 5-stage decision model that begins at the point just after an ethical issue has been identified and ends just before an action is taken. The model includes the following distinct processes (pp. 4041): 1) gain understanding develop an understanding of all facets of the decision, including related elements of the code of ethics; 2) recognize impact explore how the decision will affect others; 3) judge alternatives identify and compare alternative decisions; 4) assess other values explore values beyond those implied by the code of ethics; 5) make final decision combine codeimplied and other values to determine the appropriate course of action. As discussed earlier, Niebuhrs responsibility ethic contains four components: 1) social solidarity ongoing network of relationships and interaction among community members; 2) interpreted action the decision makers interpretation of actions preceding the decision; 3) accountability the anticipated responses by community members to the alternative courses of action;

affected parties; decisions would be made, ideally, on the basis of the stronger argument. If this process is seriously pursued, different sets of responsible behaviors may emerge and different approaches might arise among the firms with respect to AES development and implementation. The responsibility framework provides guidance for developing a context within which priorities can be set by the affected stakeholders facing a particular ethical dilemma within a specific set of circumstances. The framework does not prescribe a set of rules to be applied to all AES decisions under all circumstances. It does, however, seek to emphasize the importance of context in ethical decision making. It describes a process through which the interests of stakeholders can be explored and the conditions necessary for this process to succeed. With these limitations in mind, the following section presents an example of how the responsibility ethic can be applied in determining the ethical actions associated with AES development and implementation.

Implementation of a responsibility ethic The ethical issue in question is whether an audit firm should undertake the development and implementation of an expert system designed to make some set of significant judgments as part of the standard audit program. In this section, we draw upon a cognitive model of the ethical decision making process of auditors which was developed and validated by Lampe and Finn (1992). We combine this model with the four elements of the responsibility ethic to show how a responsibility ethic can be used as a framework for approaching ethical dilemmas. Niebuhrs work provides a general, abstract framework that can be applied in identifying and evaluating the nature of ethical issues surrounding the development and use of expert systems. The framework also allows for recognition of the pervasiveness of system-related effects throughout many levels of analysis. We begin this section by summarizing the elements of the Lampe and Finn model, which was designed to accommodate a single auditor

A Responsibility Ethic for Audit Expert Systems 4) response the decision regarding the best alternative response to the ethical dilemma and the associated action. The four components of the responsibility ethic are related to the five decision processes identified by the cognitive decision model, and a simple transformation can be used to equate the two, as shown below: Niebuhr 1) social solidarity 2) interpreted action 3) accountability 4) response Lampe and Finn 4) assess other values 1) gain understanding 2) recognize impact 3) judge alternatives 5) make final decision

345

Before the two perspectives are consolidated, it is important to note the major differences between the two. The most important difference between the responsibility approach and the cognitive process models is that the latter begins with the assumption that an ethical dilemma has already been identified. As discussed earlier, a very important shortcoming in the AES literature is its failure to treat ES implementation and use as an ethical dilemma. From a responsibility perspective, identification of ethical dilemmas must be problematized. All decisions that have the potential to significantly affect the interests of stakeholder groups have an ethical component. Thus, the process of identifying stakeholders, their related interests, and exposing potential problems should be ongoing. The second difference between Niebuhrs approach and the cognitive approaches lies in the relationship between the decision maker and other community members. The cognitive models assume that the decision maker is both thinking and acting in isolation from other stakeholders when making an ethical decision. The decision maker, working alone, is faced with the difficult task of conceptualizing the values, interests, and reactions of others to the alternative courses of action available. Niebuhrs approach views each action as only a small part of an ongoing relationship among members of a community. Through this process, stakeholders continually reveal, discuss and defend their interests. The decision maker, as a member of an ongoing

community, internalizes the interests of others, thus blurring the distinction between self and other. The interests addressed when a decision is formulated are those of both the decision maker and the stakeholders. Appropriate responses to ethical dilemmas are developed through consensus among group members rather than unilaterally by an individual decision maker working alone. To address these differences, one important change is made to the Lampe and Finn model. The fourth step, assess other values, is moved to the beginning of the process, and would actually take place before the identification of a specific ethical dilemma. This step would then provide a context for identifying and resolving dilemmas. In Niebuhrs framework, identification of ethical dilemmas is an ongoing problem for which the interests of stakeholders provide the background for the process. Niebuhr contextualizes all dilemmas by recognizing the ongoing nature of the relationships and interactions among community members. Thus, social solidarity is listed first awareness of the interests of other members of the community is an ongoing process that both precedes and contextualizes all ethical dilemmas.

Social solidarity When implementing a responsibility based approach to ethical decision making, accomplishing a sense of social solidarity requires two activities: identification of stakeholders and establishment of a communal discussion. First, the decision maker seeks to identify the relevant stakeholder groups and to understand their rights and interests. In the Lampe and Finn model, this process does not occur until a specific dilemma has been identified and the moral course of action has been determined by the decision maker. When describing this phase, Lampe and Finn attempt to allow for reasons why a decision maker might deviate from some logical course of action prescribed by the code of ethics or by the decision processes associated with the auditors level of moral development. For example, there may be an obvious technical conclusion with

346

Jesse F. Dillard and Kristi Yuthas members for actions taken. Before a response to the ethical dilemma is formulated, the decision maker explores the potential consequences of the decision for all constituents, along with their responses to the response. The accountability stage is Lampe and Finns recognize impact and judge alternatives stages. Combining these stages highlights the fact that the morality of alternatives cannot be judged solely using deontological grounds, as is implied by the Kohlberg hierarchy on which cognitive models are based. The decision maker, as part of a community, recognizes that all actions have consequences and will propagate sequences of future actions. Therefore, recognizing the impact of potential responses to the dilemma is an integral part of judging alternatives.

respect to a disclosure issue, but client relationships, personal gain, or community pressures might result in an action inconsistent with auditing norms. Under the responsibility approach, the auditor does not consider the interests of self and other stakeholders after an initial judgement is made. Instead, the process of understanding and acting upon community interests is ongoing and affects every decision and action made within that context. The responsibility component related to this decision phase is solidarity. The expert is motivated by the ethical framework to consider such things as the implications for the profession, client responsibilities, social responsibilities, demands for efficiency, consistency with professional norms, competitive pressures and social expectations with respect to technology.

Response Interpreted action This stage is comparable to Lampe and Finns recognize impact stage. Lampe and Finn conceptualize this process as obtaining data about the situation. When ethical dilemmas arise through the course of monitoring the stakeholder environment, the decision maker considers the context or history of the dilemma before acting, in an attempt to understand the context or history of the dilemma. Although the interests of other stakeholders are ongoing and continually addressed, the decision maker must think about the interactions related to the current dilemma. When an ethical dilemma is addressed by the decision maker, the decision maker must first consider the historical interactions among community members on matters related to the issue of importance. The parties to the dilemma must consider past actions and responses to those actions. The actors must also consider their own and others interpretations of those actions before forming a solution to the current dilemma. Once the decision maker and constituents have iteratively considered the alternatives and ethical issues associated with them, a response can be formulated. The response stage actually goes beyond Lampe and Finns make final decision phase, because it suggests that an action consistent with the decision will be carried out. Although Lampe and Finn do not discuss what happens after the decision is made, their work relies on Rest, who suggests that after a response to a dilemma is formulated, it will only be carried out if the decision maker has the intention, along with the perseverance and ego strength, required to do so. Again, the cognitive models assume that the decision maker is acting alone. Under the responsibility framework, constituents who might inhibit the decision makers ability to carry out the decision will have participated in the decision making process and in identification of the desired behavior. The stakeholders also recognize the ongoing nature of their relationships with the decision maker. Further, the decision maker has already considered, and eliminated, alternatives to enacting the appropriate response. Therefore, the processes of making and carrying out the decision are combined in the responsibility framework.

Accountability In the responsibility framework, the decision maker is held accountable to other community

A Responsibility Ethic for Audit Expert Systems Example of implementation of the responsibility framework In the previous section, we described the steps required for implementation of the responsibility ethic. In this section, we provide an example of what might transpire when the responsibility framework is used by an audit firm to explore the ethical contours of an AES implementation. As discussed above, the first stage of the model is social solidarity in which the audit firm constituents and their interrelationships will be recognized. This stage is ongoing, and therefore precedent to any particular ethical dilemma. The subsection below discusses the rights and responsibilities of major constituent groups likely to be affected by the AES. Once the stakeholders and their interests have been recognized, the firm begins the process of discussing the ethical dilemma (development of the AES) with its constituents. As addressed earlier, a valid discussion relies on conditions of fair discourse, in which participants can express and argue for their particular rights and interests without distorting those beliefs and without fear of retaliation. When engaging in this discussion, the participants would work through the remaining three steps of the framework interpreting past actions, exploring the consequences of alternative solutions, and producing a response to the ethical dilemma. In the final two parts of this section, we present an example of the ethical issues that might arise as part of these discussions. Although in a real decision making setting, the firm would seek reactions from all constituents, we provide examples of the interaction between the firm and the two constituent groups most strongly affected by the development and use of the AES. We examine the results of a simulated discussion between the firm and the expert or experts who are to lend expertise to the AES to be developed. We then examine the results of a simulated discussion between the firm and the owners/shareholders representative of a client firm in which the AES will be used as part of the audit program. Although the interactions exemplified below represent only a subset of those to be considered in a full implementation of the responsibility

347

approach, they are sufficient to demonstrate the importance and benefits of the responsibility framework. The examples show that by working through Niebuhrs stages, ethical issues emerge at each stage issues that vary across constituencies as well as stages. The example shows that through the responsibility-oriented process an ethical dilemma and its context can be more completely defined.

Rights and responsibilities of audit constituents The first step in the process is social solidarity. The firm becomes aware of its place in the social world by recognizing the interests of and interactions among community members. In the case of audit expert systems, members include: the expert or experts who lend expertise to the development of the system; the audit firm, represented by the controlling management group; the firms employees and users who are affected by system development and/or implementation; the audit client(s) for whom the system will be used as the basis of judgments; investors in the client firm;8 the public accounting profession; governmental and regulatory agencies; and other affected members of society. Responsible behavior requires that the relevant issue set be derived from the perspective of each community member in relation to each other member. The interaction set is illustrated in Figure 1. Only through examining the complexity of the context and relationships surrounding an issue can responsible behavior be determined. Niebuhrs responsibility ethic assumes a ready stock of knowledge comprised of previous behaviors and responses that, in effect, construct the frame of reference used for understanding and interpretation. An agent draws on this stock of knowledge in considering a response and in anticipating the subsequent responses of communal agents. In addressing an issue or question, not only may the agent be granted certain rights9 by the community, but also the social/ ethical fabric represented in the agents stock of knowledge suggests certain responsibilities coinciding with those rights. The following discus-

348

EXPERT EXPERT Right to expertise

EMPLOYEE/ FIRM USER Competence Professional integrity Solidarity Right to technically competent expert system Competence Professional integrity Loyalty Competency Professional integrity Loyalty

CLIENT MGT. Competence Professional integrity Competence Professional integrity Efficient and effective performance Competence Professional integrity

CLIENT Competence Professional integrity Competence Professional integrity Efficient and effective performance Competence Professional integrity Certify FS for accuracy Provide stewardship information Right to efficient and effective audit Maintain professional standards

INVESTORS Competence Professional integrity Membership Competence Professional integrity Membership

PROFESSION SOCIETY REGULATORS Competence Professional integrity Competency Professional integrity Citizenship Competence Professional integrity Citizenship Competence Professional integrity

EMPLOYEE/ Competence USER Professional integrity Respect for expertise Solidarity FIRM Fair compensation Good work environment Accept experts judgement

Fair compensation Good work environment Provide access and cooperation Recognize competence and professional integrity Maintain professional standards Recognize expertise Recognize competence and professional integrity Recognize competence and professional integrity

Right to develop and implement expert systems Accept experts judgement Compensate firm fairly Recognize competence and professional integrity Maintain professional standards Recognize expertise Recognize competence and professional integrity Recognize competence and professional integrity

Competence Professional integrity

Competence Professional integrity Certify FS for accuracy Provide conformance information Provide conformance information

Competence Professional integrity Abide by laws and regulations Pursue socially responsible actions Pursue socially responsible actions

Jesse F. Dillard and Kristi Yuthas

CLIENT MANAGEMENT

Right to efficient and effective audit Provide competent oversight Recognize management expertise Maintain professional standards

Acknowledge services provided Acknowledge services provided

CLIENT Recognize INVESTORS competence and professional integrity PROFESSION Maintain professional standards Recognize expertise Recognize competence and professional integrity Recognize competence and professional integrity

Right to comProvide petence and constructive professional input integrity including ES development Provide efficient and effective oversight Right to valid information

Pursue socially responsible actions Provide efficient and efficient and oversight

REGULATORS

Provide efficient and effective oversight Recognize management expertise Recognize management expertise

Provide efficient and effective oversight

SOCIETY

Recognize investor expertise

Acknowledge services provided

Acknowledge services provided

Right to valid information

Figure 1. Rights and responsibilities of AES constituents.

A Responsibility Ethic for Audit Expert Systems sion identifies eight areas of rights and responsibilities that may be associated with communal agents for an audit expert system decision. The expert presumes the right to his/her expertise, but given that right has responsibilities to all community constituencies. The primary responsibility to all constituencies is that of competence and professional integrity. In addition, there are responsibilities of solidarity with other firm employees/users, of loyalty to the firm, of membership to the profession, and citizenship to the community. The firm employee/user presumes the right to a technically competent expert system, but given that right has a primary responsibility, to all constituencies, of competence and professional integrity. In addition, there are responsibilities of respect toward and solidarity with the expert, loyalty to the firm, effective and efficient performance to the firm and the client, membership to the profession, and citizenship to the community. The firm presumes the right to develop and implement an audit expert system, but given that right has responsibilities to all community constituencies. The responsibilities to the expert and the employee/user are fair compensation, good work environment, security, training, trust-worthiness, and respect. The firms responsibilities to the client, profession and society are competence, professional integrity, and effective and efficient service. The firm is responsible to investors and regulators in certifying the accuracy of the information provided by the client and to society in abiding by its laws and regulations. The client presumes the right to an effective and efficient audit, but given that right has the following responsibilities: accept the experts expertise and judgment, provide access and cooperation to employee/user, compensate the firm fairly for its services, provide information reflecting stewardship of resources and results of actions to investors, acknowledge the services provided by the profession, provide information indicating level of conformance to regulations, and provide goods and services in a socially responsible way within the laws and regulations of society. The clients investors have a right to an effec-

349

tive and efficient audit, but given that right have the following responsibilities: recognize the competence and professional integrity of the expert, employee/users and the firm; provide professional and competent oversight of client activities and recognize managements competence and professional expertise; acknowledge the service provided by the profession; provide information indicating the level of conformity to regulations; and carry out resource allocation decisions effectively and efficiently in a socially responsible way within the laws and regulations of society. The profession presumes the right to expect competence and professional integrity from those involved in expert system development and implementation, but given that right has a responsibility to all constituencies to maintain professional standards. In addition, there are responsibilities to recognize the expertise of member professionals, to provide constructive input to the regulatory process, and to carry out its obligations in a socially responsible manner within the laws and regulations of society. The regulators presume the right to valid information, but given that right have the responsibility to recognize the competence and professional integrity of the expert, the employee/user and the firm. The regulators have a responsibility to the other constituencies to provide effective and efficient regulatory oversight as well as to acknowledge the client managements competence and professional expertise. Society presumes the right to valid information but given that right has the responsibility to recognize the competence and professional integrity of the experts, employees/users, and the firm. Further, society has a responsibility to recognize the competence and professional expertise of the client and the investors as well as their position as allocators of capital resources. Society also has the responsibility to acknowledge the service provided by the profession and the regulatory groups.

The expert/firm relationship Applying Niebuhrs framework requires that, in addition to the ongoing social context, past and

350

Jesse F. Dillard and Kristi Yuthas sion have been organized into six categories: commitment, quality of work, quality of the audit, quality of the system, quality of work life, and competitive pressures. Examples of the issues that arise with respect to commitment include: does the firm have an obligation to increase its commitment to the expert? does the expert have an obligation of increased commitment to the firm? will the experts status within the firm increase? will exposure of expertise negatively affect the experts status? will the firm accept liability for system failures? will the firm use the system for the intended purpose? will the firm inform the expert of the systems use? will each party assist in the upgrading of the system? will the expert sell expertise to competitors? will the expert leave the firm prematurely? will the expert support the use of the system? Examples of issues that arise with respect to the nature of the work include: will the nature of the experts work be significantly changed? will the system affect the work of other firm employees? will auditors be required to use the system? can the expert withdraw from the system development project? can the expert delegate tasks as a result of developing the system? can the expert free time for more demanding tasks? can the work of nonexperts be upgraded? Examples of issues that arise with respect to the quality of the audit include: will the system improve the quality of the audit through access to expertise? will the system provide concretization and standardization of judgment? will the expert conscientiously participate in system development and implementation? will the expert misrepresent knowledge? will the expert identify knowledge gaps? will the firm continue to research potential negative system consequences? will the firm withdraw the system if audit quality is degraded? will the firm withdraw the system if the expert-client relationship is degraded? Examples of issues that arise with respect to the quality of the expert system include: will the participants in the system development be competent and adequately trained? will the system be technologically accurate? will the system be able to be used effectively and efficiently? will the

expected future actions be considered when determining responsible action. The first illustration of the frameworks application is based on a scenario in which an expert and a firm consider the development and implementation of an audit expert system. The parties are assumed to have enjoyed an ongoing positive relationship characterized by loyalty and trust. The scenario developed here is designed to demonstrate that even within this narrow, cooperative context, a large number of ethical issues arise that, to this point, have not been recognized or explored in the AES literature. Figure 2 lays out the basic elements of Niebuhrs framework from the perspectives of the expert and firm. It identifies some of the many issues that might arise in a dialogue regarding system development and implementation. The list of issues is not intended to be exhaustive. Its purpose is to illustrate the breadth and complexity of the problem and to make explicit the general responsibilities described above. The list demonstrates that a full exploration of the relevant issues extends far beyond the previously addressed problems of ownership and liability. The firm/expert figure presents an example of issues that might emerge from discourse between these constituent groups. Identifying such issues in real-life audit settings would necessitate a discourse in which each partys position and concerns are presented and debated. In determining an ethical course of action with regard to the development and use of an ES, the primary consideration for the expert and firm is their accountability to other constituent groups affected by the development of the system. The action and social solidarity portions of Figure 2 represent the participants perceptions of previous actions and relationships between the parties. The accountability portion of the dialogue is critical to the determination of an ethical course of action, because it represents anticipated effects of ES development and use. Thus, parties must thoroughly explore their responsibilities toward each other when selecting a course of action. Numerous responsibilities are likely to arise through such analysis, and a number of these are represented in the accountability portion of Figure 2. Responsibilities relating to this discus-

A Responsibility Ethic for Audit Expert Systems system be updated to accommodate changing circumstances? will the system be used appropriately and in appropriate contexts? will the system be adequately tested? will the firm and users be fully aware of technological limitations? Examples of issues that arise with respect to the quality of work life include: will the expert be able to delegate decisions to nonexperts, freeing time for more challenging assignments? will nonexperts receive additional authority and responsibility as a result of having access to the system? will nonexperts be provided an opportunity to develop expertise? will the system routinize work, limit creativity and learning? will participation create a time burden? will participation create unnecessary frustration? will continued participation be at the discretion of the expert and other participants? will the firm withdraw the system if the quality of work life is degraded? Examples of issues that arise with respect to competitive pressures include: will the development of the expert system help relieve competitive pressures? will the system reduce audit costs by shifting tasks to lower cost, less expert, personnel? will the system reduce the number of high-priced experts needed to be retained? will the value of expertise be downgraded by participation? will liability be reduced through standardizing some audit judgments? will liability be reduced through an increased ability to substantiate decision processes? will the system be used according to industry norms?

351

The firm/investor relationship Figure 3 lays out the basic elements of Niebuhrs framework from the perspectives of the audit firm and the investors in the client firm for whose audit the proposed expert system will be used. This scenario identifies some of the many issues that might arise in a dialogue between the two regarding the development and implementation of such a system. Again, the list is not comprehensive, but illustrates the breadth and complexity of the problem and the need to extend consideration of ethical issues beyond the expert/firm relationship that has been the primary focus in previous research.

The responsibilities identified in this scenario revolve around issues of the effectiveness and efficiency of audit services, expertise, competence, and professional integrity as part of the process of attesting to the accuracy of the information supplied to the investors. This figure illustrates all four components of Niebuhrs responsibility ethic. In contemplating the response (the development and use of the ES), the firm and investors must maintain accountability they must explore the potential effects of the system on other constituent groups. Two categories of issues are likely to emerge from such an endeavor: issues concerning audit effectiveness and efficiency, and issues concerning expertise, competence, and professional integrity. The issues concerning audit effectiveness and efficiency are: Will the system improve the overall effectiveness of the audit? Will the system increase the overall efficiency of the audit? Will the system increase the productivity of cheaper nonexperts? Will the system reduce the time required to perform tasks? Will the system reduce the need for highly compensated experts? Will the system improve the consistency of the audit process? Will the system improve the consistency of audit work papers? Will the system reduce audit risk? Will the system reduce the likelihood of litigation? Will the system contribute to high quality audits? Will the system contribute to reasonably priced audits? Will the system facilitate meeting industrial quality standards? Will the system affect billing rates? Will the system enhance the ongoing professional relationship between the firm and investors? The issues concerning expertise, competence, and professional integrity include: Will the number of experts working on the audit be reduced? Will system development rely on qualified personnel? Will the system increase the development of expertise by audit personnel? Will the system be adequately validated? Will the system be used in place of experts only when appropriate? Can the system be appropriately applied to the clients particular circumstances? Will system judgments be validated by a human expert? Will system judgments be questioned by users? Will decisions be backed by clear logic trails? Is the system technologically justified? Will the system facilitate meeting professional standards?

RESPONSIBILITIES OF EXPERT AND FIRM From perspective of: ACTION Commitment Firm has been committed to all employees Firm has not jeopardized job status of employees Firm has supported and encouraged development of expertise Firm has rewarded expertise Prior changes in work have been necessary Work is similar to that in other firms Previous technical innovations have not significantly affected work Firm has maintained an acceptable level of audit quality in the past Previous technological innovations have not degraded audit quality Firm has thoroughly explored consequences of ES Firm has expected manageable work load in the past Firm has treated employees with respect Firm has provided challenging work environment Firm has provided a quality work environment for all employees Firm has made technical adjustments to address increasing competition Salaries have been adjusted according to market demands Firm can get another expert if one opts not to participate Firm could fire employee Expert has tenure Expert has shown continued commitment to firm Expert has been willing to share expertise with other members of firm Expert has been compensated for knowledge attained on the job Past changes have been accepted by expert Expert understands need for ongoing change to keep pace with industry EXPERT FIRM

352

Nature of work

Jesse F. Dillard and Kristi Yuthas

Quality of audit

Expert has performed competently in prior audits Expert has mastered changing audit requirements Expert has adopted technological innovations supporting the audit Expert has been satisfied with work requirements Expert has adapted to challenges in job and changing audit environment Expert has accepted quality of work life associated with all positions in the firm Expert has accepted reasonable monetary benefits Expert has accepted changes necessitated by increasing market pressures from clients and firms to economize

Quality of work life

Competitive pressures

ACCOUNTABILITY Commitment Participation will strengthen firms commitment to ongoing relationship System will boost experts status in the firm Knowledge gaps or judgements that are incongruous with other experts wont negatively affect status Firm will still be liable for system audit failures Firm will still use AES only for stated purposes Firm will inform expert of uses of system Selecting expert for project will boost experts morale and increase commitment to firm Expert will help upgrade the system in the future Expert will not sell knowledge to competitors Expert will not quit prematurely Expert will support use of the system Expert will maintain confidentiality with regard to system

Nature of work

Participation will not significantly change the nature of expert work System will reduce or enhance work for other employees Expert will not be required to use system Expert will be able to withdraw from system project if uncomfortable with procedures or results Participation will provide basis for improving the quality of the audit Firm will continue to research and address potential negative systems consequences Firm will withdraw system if quality is degraded Firm will withdraw system if expert/client relationship is degraded Firm will develop high quality, reliable, usable system System will be used appropriately and in appropriate contexts System will be tested adequately Firm is fully aware of technological limitations System developers are adequately trained Participation will enhance quality of experts work Expert will have guidance in performing the audit Other employees will receive greater challenges and responsibilities, as well as greater opportunities for learning Use will not routinize work, limit creativity or learning Participation wont create time burden or unnecessary frustration Continued participation will be at the discretion of the expert Participation will help relieve competitive pressures facing the firm Value of expertise will not be downgraded by participation

Experts work will not change significantly ES related aspects of expert work may be delegated Experts time will be freed for ever more demanding decisions Nonexpert work and capabilities will be upgraded

Quality of audit

System will enhance quality of audit through access to multiple experts System will provide solidification and standardization of judgements Expert will participate conscientiously and will not withhold or misrepresent knowledge Expert will identify gaps in knowledge Participants in ES development will be competent ES will be technically accurate ES will be used appropriately Experts and other users will be able to use ES effectively ES will be upgraded to accommodate changing circumstances ES is expected to enhance quality of work life for experts who can push down some decisions to nonexperts, freeing time for other challenges Nonexperts will receive more responsibility and opportunity to develop expertise more rapidly

A Responsibility Ethic for Audit Expert Systems

Quality of expert system

Quality of work life

Competitive pressures

ES will reduce cost of audit by shifting decisions downward ES may reduce liability through standardization of some audit judgements, making visible (and subject to evaluation) the judgement process Use of ES will be in accordance with industry norms

RESPONSE Expert agrees to lend expertise to AES Expert shows ongoing commitment to success of firm and to audit quality Expert accepts changes to work that will arise as a result of the system Firm enlists experts assistance in developing expertise Firm expresses confidence in experts ability Firm expresses confidence that ES technology will improve the audit

353

Figure 2. Mutual responsibilities of the expert and firm.

354

RESPONSIBILITIES OF FIRM AND CLIENTS INVESTORS From perspective of: ACTION Contractual relationship Investors have been satisfied with previous years audits Investors benefit from ongoing relationship Repeat audits have been profitable in the past Investors have expected high quality audits Investors have procedural, Investors have Investors have Investors have allowed the firm control over personnel, and technological decisions accepted level of expertise exhibited accepted billing levels accepted the use of trainees for some tasks Firm has performed satisfactorily on prior audits Firm benefits from ongoing relationship Previous repeat audits have been reasonably priced Firm has met industry quality standards Firm has used well-trained personnel and appropriate procedures and technology Firm has used personnel with appropriate levels of service Firm has used billing levels appropriate to task Use of trainees has not impeded audit quality or efficiency Firm has commitment to use up-to-date procedures Firm has used techniques comparable to those used by peer firms Firm has used appropriate audit procedures Firm has kept up with technological advancements Firm has used technology comparable to that used by peers Firm has used appropriate technology to perform previous audits FIRM CLIENTS INVESTORS

Jesse F. Dillard and Kristi Yuthas

Personnel

Procedures

Investors have expected use of up-to-date audit procedures Investors have expected use of procedures comparable to those used by peer firms Investors have accepted firms choice of audit procedures Investors have expected continued research into the use of audit-enhancing technologies Investors have expected firm to use technology comparable to that used by peer firms Investors have accepted the use of advanced audit technology

Technology

Competitive pressures

Investors understand competitive pressures in the audit profession Investors understand the need for the use of productivityenhancing tools Investors understand pressure to use nonexperts when possible

Firm faces strong competition from peer firms Firm must use productivity-enhancing techniques to remain competitive Firm will use cheapest personnel appropriate to the task

ACCOUNTABILITY Efficiency and effectiveness ES ES ES ES ES ES ES ES ES will enhance the overall effectiveness of the audit will enhance the overall efficiency of the audit will enhance the productivity of cheaper nonexperts will enhance the consistency of the audit process will enhance the consistency of audit work papers will reduce the likelihood of litigation will contribute to high quality audits will not affect billing rates will improve the ongoing professional relationship between the expert and firm ES ES ES ES ES ES ES ES will increase the overall effectiveness of the audit will increase the overall efficiency of the audit will reduce the need for highly compensated experts will reduce the time required to perform tasks will reduce audit risk and likelihood of litigation will contribute to reasonably priced audits will facilitate meeting industry quality standards will improve the ongoing professional relationship between the expert and firm

A Responsibility Ethic for Audit Expert Systems

Expertise, competence and professional integrity

ES will enhance the development of expertise by audit personnel ES judgments will be questioned by the user Decisions will be backed by clear logic trails ES is technologically justified ES will facilitate meeting professional standards

System development will use qualified personnel Number of experts working on the audit will be reduced ES will be adequately validated ES will be appropriately applied ES will be used by competent personnel ES judgments will be validated by a human expert

SOCIAL SOLIDARITY Firm Firm Firm Firm and and and and investors investors investors investors are mutual participants in the audit have mutual concern for the quality of the audit have mutual concern for the efficiency of the audit seek to avoid legal liability

Figure 3. Mutual responsibilities of the firm and the clients investors.

355

356 Concluding remarks

Jesse F. Dillard and Kristi Yuthas level of loyalty and trust among the participants, and their willingness to engage in legitimate and comprehensive dialogue. The framework developed is intended to be prescriptive as opposed to predictive and can be applied, with slight modifications, to any of the stakeholder groups associated with the development and implementation of AES. In addition, implementation of the framework facilitates the specification of behavioral propositions. First, such an analysis would allow prescriptive processes to be identified. Decision makers could be observed to determine how closely their actual decision processes adhere to the model. Second, the responsibility ethic provides an indication of relevant issues that should be evaluated at each phase for the decision process in order for the undertaking to be considered an ethical one. The extent to which such issues are considered by a decision maker might indicate some level of ethical sensitivity. The extent to which these dimensions are identified beforehand might prevent decision makers from excluding important factors when considering alternatives. Third, the extent to which the requisite dialogical characteristics of generalizability, autonomous evaluation, role taking, power neutrality and transparency have been obtained at each decision phase can be employed as a validity metric. Fourth, and more generally, the responsibility ethic provides a rich context within which to consider decisions having ethical implications. Not only does it extend beyond the professional code implied model (AICPA, 1988), it also provides a theory-grounded context within which cognitive decision processes are carried out. Fifth, at an operational level, by using the proposed framework it may be possible to identify specific elements relevant to explaining the degree to which firms incorporate ethical considerations into their decision processes and the extent to which formal procedures are in place to do so. As a result, explanations of differences in the extent and timing of AES applications across audit firms could be identified. Though these propositions could be reformulated as testable hypotheses, we believe that at this stage such implied specificity is unwarranted and unduly limits the scope of the discussion. The

Previous research on the ethical development and use of expert systems in auditing has focused on a limited set of ethical issues. This paper has demonstrated how Niebuhrs theory of the responsible self can be used both to broaden the concept of what constitutes an ethical issue and to provide a framework for identifying what constitutes responsible or ethical behavior. Essentially, behavior is responsible if it takes into consideration the previous acts and relationships of all parties affected by the act and anticipates the consequences of the act on these parties as well as their expected reactions. The framework compels those involved in developing expert systems to consider how their actions fit into the context of ongoing relationships not only among experts and firms, but also all other affected parties. The preceding discussion has described discourse requirements that provide the means and conditions necessary for implementing Niebuhrs ideas. To achieve consensus regarding the ethical development and implementation of an expert system, the following conditions should be sought: all participants should have mutual consensus as a goal; there should be free and equal access to the discussion; and all participants should be appropriately informed, rational, and uncoerced. Although it is unlikely that these conditions can be fully attained in real-life settings, they provide an ideal toward which the discussion can aspire and through which discussions can be evaluated. The responsibility ethic arises as a result of the trust and loyalty necessary for the sincere consideration of issues (behaviors and responses) through ongoing communal discussion. As the analysis progresses, the parties may evolve toward a utilitarian resolution of some issues and a contractarian solution for others, but all will be considered and carried out with an attitude of responsibility motivated by accountability through anticipated responses with the ongoing community or society. The resolution of each of the issues identified in the presentation above will be situation-specific and dependent upon the agents perceptions based on past experience, the

A Responsibility Ethic for Audit Expert Systems idea that scientists ought to be concerned only with prediction, explanation and the construction of testable theories is an unacceptable corollary of the value-neutral thesis [of science] (Gaa, 1996, p. 15). Ethics research has an agenda to encourage and facilitate decision makers to act ethically. Ultimately, our own (value-laden) purpose is to aid audit firms and their stakeholders in acting more ethically when AES are developed and used. The discussion undertaken here has initiated a preliminary phase in the study of ethical issues surrounding expert system applications. It has demonstrated that deliberation can, and must, be broadened beyond its current focus on economic claims (in this case rights to intellectual property), and dialogically considered in light of a responsible self. Such discussion must consider not only the appropriate action within the context of individual rights and responsibilities, but must also give recognition to the continuing human community a community wherein agents are accountable for the consequences of their actions and in which the self and the community are strongly interlinked and continually redefined. To pursue ethical actions effectively, the audit community must understand that the fundamental form of human association is not the social contract into which persons enter as atomic individuals, making partial commitments to each other for the sake of gaining limited common ends or of satisfying certain laws. The fundamental form of human association is, rather, the face-to-face community in which ongoing commitments are the rule and in which aspects of every individuals existence are conditioned by continuing membership. Notes
1

357

Though they only briefly refer to expert systems, Mason et al. (1995) and Spinello (1995) present a discussion of the ethical issues surrounding information technology and management in general. 2 Although we use the singular form of expert, systems are often developed and tested using more than one expert. 3 As MacIntyre (1984) notes, all three perspectives are evaluated within the current modern economic

order which contains the central features of individualism, acquisitiveness, and the elevation of the market place to a central position. 4 Agent is used in the sociological sense to mean one who has the ability to act or influence, not in the more restricted neoclassical economics sense of principal-agent relationships. 5 The ideas formulated by Niebuhr are grounded in, and are evolving out of, the linguistic turn in modern philosophy. Habermas (1984, 1987), among many others, has recognized the power of language and speech acts in human reasoning and understanding. The current discussion is an integration of Niebuhrs ideas about ethics being grounded in an ongoing community and Habermas more recent ideas about the centrality of speech acts to communication and understanding and the universality of validity claims which provide the basis for human understanding and systems integration. Ethical claims must be grounded to be justifiable. Generally, these grounds are either the result of some covering law, such as Holy Scripture, or are, according to our formulation, the result of communal dialogue and interaction (Goffman, 1974), thus the centrality of the dialogical model. 6 Several variations on stakeholder theory exist (Donaldson and Preston, 1995), including descriptive stakeholder theory which suggests that firms survive by responding to their stakeholders, and instrumental stakeholder theory which suggests that firms can manage stakeholders to accomplish instrumental goals. We use normative stakeholder theory which suggests that firms are morally obligated to address the interests of stakeholders. 7 In the stakeholder literature (Carroll, 1996), an approach called issues management has been developed whereby a firm continuously scans its environment to develop an awareness of ongoing stakeholder interests and reactions to previous actions. Although this approach is certainly superior to the standard approach of ethical research, in which the agent merely thinks about the consequences of action, it does not go far enough in seeking active participation among affected stakeholders when an ethical dilemma is addressed. 8 We recognize that the clients stakeholder set goes beyond creditors and stockholders to include employees, suppliers, customers, society, etc. To limit the scope of the discussion, we assume that the interests of these stakeholders are reflected by the client criterion of continuing operations or by those of society at large. 9 These rights and responsibilities are not the result

358

Jesse F. Dillard and Kristi Yuthas

and Implications, Academy of Management Journal 20, 6591. Dungan, C.: 1983, A Model of an Audit Judgment in the Form of an Expert System. Unpublished dissertation, University of Illinois. Eining, M. and P. Dorr: 1991, The Impact of Expert System Usage on Experiential Learning in an Auditing Setting, Journal of Information Systems (Spring), 116. Gaa, J.: 1996, Ethics Research and Research Ethics, Behavioral Research in Accounting 8 (Supplement), 1224. Gal, G. and P. Steinbart: 1992, Interface Style and Training Task Difficulty as Determinants of Effective Computer-assisted Knowledge Transfer, Decision Sciences ( January-February), 128143. Goffman, E.: 1974, Frame Analysis (Northeastern University Press, Boston). Graham, L., J. Damens and G. Van Ness: 1991, Developing RISK ADVISOR: An Expert System for Risk Identification, Auditing: A Journal of Practice and Theory 10 (Spring), 6996. Habermas, J.: 1984, 1987, The Theory of Communicative Action, Translation by T. McCarthy (Beacon Press, Boston). Hansen, J. and W. Messier: 1986. A Preliminary Investigation of EDP-EXPERT, Auditing: A Journal of Practice and Theory (Fall), 109123. Kelley, K.: 1984, Expert Problem Solving for the Audit Planning Process. Unpublished dissertation, University of Pittsburgh. Kettner, M.: 1993, Scientific Knowledge, Discourse Ethics, and Consensus Formation in the Public Domain, in E. Winkler and J. Coombs (eds.), Applied Ethics (Blackwell Publishers, Cambridge, MA), pp. 2845. Khalil, O.: 1993, Artificial Decision-Making and Artificial Ethics: A Management Concern, Journal of Business Ethics, 313321. Khani, P. and S. Zarolwin: 1995, A Journal Survey: Counting on Technology, Journal of Accountancy (May), 5968. Lampe, J. and D. Finn: 1992, A Model of Auditors Ethical Decision Processes, Auditing: A Journal of Practice and Theory 11 (Supplement), 3359. MacIntyre, A.: 1984, After Virtue, 2nd ed. (University of Notre Dame Press, Notre Dame, IN). Mason, R., F. Mason and M. Culnan: 1995, Ethics of Information Management (Sage Publishers, Thousand Oaks, CA). Meservy, R., E. Denna and J. Hansen, 1992, Applications of Artificial Intelligence to Accounting, Tax, Audit Services: Research at

of some natural law or covering law external to the community. They are the result of ongoing relationships and discursive activities of the community which have resulted in agreed upon values and behavioral norms.

References
AICPA: 1988, Code of Professional Conduct (American Institute of Certified Public Accountants, New York). Bailey, A., K. Han and A. Winston: 1988, Technology, Competition and the Future of Auditing, in A. Bailey (ed.), Auditor Productivity in the Year 2000 (Council of Arthur Young Professors, Reston, VA), pp. 2350. Baldwin-Morgan, A.: 1998, Expert Systems for Audit Tasks Applications and Impacts, in M. Vasarhelyi and A. Kogan (eds.), Artificial Intelligence in Accounting and Auditing, Volume 4 (Markus Wiener Publisher, New York), pp. 99-110. Baldwin-Morgan, A.: 1993, The Impact of Expert System Audit Tools on Auditing Firms in the Year 2001: A Delphi Investigation, Journal of Information Systems (Spring), 1634. Broderick, J.: 1988, A Practical Decision Support system. In in A. Bailey (ed.), Auditor Productivity in the Year 2000 (Council of Arthur Young Professors, Reston, VA), pp. 131148. Brown, C.: 1991, Expert Systems in Public Accounting: Current Practice and Future Directions, Expert Systems with Applications 3, 318. Brown, C. and D. Murphy: 1990, The Use of Auditing Expert Systems in Public Accounting, Journal of Information Systems (Fall), 6572. Carroll, A.: 1996, Business and Society 3rd ed. (Southwestern College Publishing, Cincinnati, OH). Dillard, J. and R. Bricker: 1992, A Critique of Knowledge-based Systems in Auditing: The Systematic Encroachment of Technical Consciousness, Critical Perspectives on Accounting 3 (3, September), 205244. Dillard, J. and J. Mutchler: 1988, Knowledge-based Expert Systems in Auditing, in C. Ernst (ed.), Management Expert Systems (Addison-Wesley). Dillard, J. and K. Yuthas: 1997, Ethical issues in Expert Systems: Lessons for Moral Philosophy, Research on Accounting Ethics 3, 99117. Donaldson, T. and L. Preston: 1995, The Stakeholder Theory of the Corporation: Concepts, Evidence,

A Responsibility Ethic for Audit Expert Systems

Brigham Young University, Expert Systems with Applications, 213218. Mitroff, I. and R. O. Mason: 1989, Deep Ethical Issues in the Design of Information Systems, Expert Systems Review (Fall), 2125. Murdock, I.: 1992, Metaphysics as a Guide to Morals (Penguin Books, New York). Niebuhr, R.: 1963, The Responsible Self (Harper Press, San Francisco). Nozick, R.: 1974, Anarchy, State, Utopia. OLeary, D. and P. Watkins: 1989, Review of Expert Systems in Auditing, Expert Systems Review (Spring-Summer), 322. Pei. B. and J. Reneau: 1990, The Effects of Memory Structure on Using Rule-based Expert Systems for Training: A Framework and an Empirical Test, Decision Sciences 21, 263286. Pei, B., P. Steinbart and J. Reneau: 1994, The Effects of Judgment Strategy and Prompting on Using Rule-based Expert Systems for Knowledge Transfer, Journal of Information Systems (Spring), 2142. Rawls, J.: 1971, A Theory of Justice (Belknap Press, Cambridge, MA). Rest, J.: 1986, Moral Development: Advances in Research and Theory (Preager Publishers, NY). Specht, L., R. Trotter, R. Young and S. Sutton: 1991, The Public Accounting Litigation Wars: Will Expert Systems Lead the Next Assault?, Jurimetrics: Journal of Law, Science, and Technology (Winter), 247257. Spinello, R.: 1995, Ethical Aspects of Information Technology (Prentice-Hall, Englewood Cliffs, NJ). Srivastava, R., P. Shenoy and G. Schafer: 1990, Belief Propagation in Networks for Auditing, Proceedings of the 1990 U.S.C./Deloitte and Touche Audit Symposium. Steinbart, P.: 1984, The Construction of an Expert System to Maker Materiality Judgments. Unpublished dissertation, Michigan State University. Steinbart, P.: 1987, The Construction of a Rulebased Expert System as a Method for Studying Materiality Judgments, The Accounting Review ( January), 97116.

359

Steinbart, P. and W. Accola: 1994, The Effects of Explanation Type and User Involvement on Learning from and Satisfaction with Expert Systems, Journal of Information Systems 8 (Spring), 117. Sutton, S. and J. R. Byington: 1993, An Analysis of Ethical and Epistemological Issues in the Development and Implementation of Audit Expert Systems, Advances in Public Interest Accounting 5, 231244. Sutton, S., V. Arnold, and T. Arnold: 1995, Toward an Understanding of the Philosophical Foundations for Ethical Development of Audit Expert Systems, Research on Accounting Ethics 1, 6174. Tomas, A., Expert Systems Applications in Accounting, Baldwin-Morgan, A.: 1998, Expert Systems for Audit Tasks Applications and Impacts, in M. Vasarhelyi and A. Kogan (eds.), Artificial Intelligence in Accounting and Auditing, Volume 4 (Markus Wiener Publisher, New York), pp. 728. Vasarhelyi, M. and A. Kogan (eds.): 1998, Expert Systems for Audit Tasks Applications and Impacts, in Artificial Intelligence in Accounting and Auditing, Volume 4 (Markus Wiener Publisher, New York). Willingham J. and G. Ribar: 1988, Development of an Expert Audit System for Loan Loss Evaluation., in A. Bailey (ed.), Auditor Productivity in the Year 2000 (Council of Arthur Young Professors, Reston, VA), pp. 171186. D. Yang and M. Vasarhelyi: 1998, Expert Systems for Audit Tasks Applications and Impacts, in M. Vasarhelyi and A. Kogan (eds.), Artificial Intelligence in Accounting and Auditing, Volume 4 (Markus Wiener Publisher, New York), pp. 2946.

Anderson School of Management, University of New Mexico, Albuquerque, NM 87131, U.S.A. E-mail: dillard@unm.edu yuthas@anderson.unm.edu