The future of expert systems.
Expert systems are a tool to aid management in decision making. Today, expert systems are not common in accounting practices. Professional accountants are still responsible for making most decisions, even the most mundane and routine. The time is right for expert systems in these firms.
Expert System Defined
Expert systems are the most mature and widely-used commercial field of artificial intelligence. In an expert system, the computer applies reasoning methodologies in a knowledge-specific domain to render advice or make recommendations, much like a human expert. Expert systems achieve high levels of performance in task areas that, for human beings, require years of special education and training.
Expert System Components
Expert system components, summarized in Exhibit 1, include: a knowledge base, an inference engine, a user interface and an explanation facility. Tax accrual--the process of identifying tax-book differences--is a prime area for expert system application in which the components can easily be identified. ExperTax, a tax accrual expert system developed by Coopers and Lybrand to assist in analyzing tax accruals, will be used throughout to facilitate explanation.
The first component, the knowledge base, is a collection of facts, rules and procedures organized into schemas. It is the assembly of all information and knowledge specific to the area of accounting in which the expert system is applied. ExperTax stores information in frames. Information attached to each frame relates to how and when to use the frame, to determine what should happen next, and to determine what to display or print. Most of the frames include precondition rules, possible answer rules, marginal note instructions and "why" messages. For example, the system will ask what a client's bad debt write-off method is and depending on the method chosen, will ask a series of questions until it reaches a final tax accrual answer. The knowledge base contains all the information pertaining to each method of write-off as well as questions, answers and instructions. Developers created the knowledge base by capturing appropriate information from human experts and other tax sources such as the Internal Revenue Code, court cases and revenue rulings.
The second component, the inference engine, performs the reasoning function in the expert system. The inference engine is the "brain" of the expert system and is the component that processes information. The engine recognizes the most appropriate rule from the knowledge base and matches it with information supplied by the user. The inference engine links the rules in the knowledge base to derive a conclusion and communicates this conclusion to the user through the interface.
The third component, the user interface, controls user interaction, accepting commands from the computer keyboard and displaying the results generated by the inference engine. When performing this function, the user interface employs three horizontal windows. The top window displays area of analysis, the middle window displays questions and answers while the lower window provides "why" messages. A series of menus further aids the communication process between the user and computer.
The final component, the explanation facility, explains the system's reasoning and justifies its conclusions. With each answer, ExperTax provides explanations such as Code Section, Treasury Regulation, Court Case, Revenue Ruling or IRS Notice.
Applicability of Expert Systems
The development of an expert system begins with the selection of possible application areas. Four prerequisites to success are:
* the existence of a human expert,
* clear definition of system/application boundaries,
* clear definitions of facts and decision rules, and
* knowledge that payback will not be immediate.
First, a human expert must exist as a source of undocumented knowledge. Other sources such as textbooks, journals, databases and working papers can be used to capture much information, but the expert system performs its greatest achievements by capturing and using the undocumented knowledge which can only be found in human minds.
Secondly, the application must have clearly defined boundaries. Objectives and limitations must be identifiable. If the scope of the project cannot be defined, appropriate information cannot be captured. An expert system cannot provide correct information for queries beyond its programmed limits.
The third prerequisite requires clear definition of facts and decision rules. Without well-defined facts and decision rules, the expert system cannot reference appropriate information for answers to queries. A human expert's performance suffers when given vague and nebulous facts. In the same perspective, the expert system will not be applicable with vague and nebulous facts because it does not have the capability of interpretation, further research or educated guessing. An expert system is only as accurate as the information captured in the knowledge base.
Finally, payback will not be immediate. Expert systems requite initial investment that is returned over an extended period of time. In most instances, substantial amounts of money and time are necessary prior to initial implementation. At that time, tangible and intangible benefits begin to accrue. Even with the most successful expert system, immediate payback does not occur with initial implementation.
Development of Expert Systems
If prerequisites are met and applicability of an expert system appears likely, three development approaches are available. They include custom development of the expert system, purchase of a fully developed expert system, or purchase of a package which requires creation of a knowledge base.
The first approach entails custom development of an entire expert system internally by an accounting firm. This approach is costly and time consuming. Big firms are more likely than small firms to custom develop expert systems because of the significant resources involved.
The second approach for developing an expert system is buying a fully developed expert system for a specific application. Few packages of this type are available because incorrect system results could lead to legal action against the developer. Also, developers have tended to protect the expertise developed to maintain differential advantage. If an accountant does find a fully developed expert system package, it rarely produces exactly what is needed or wanted, only something close to what is specified.
The third approach for developing an expert system is buying a rule-based package and creating the knowledge base. This involves purchasing a basic shell and having the accountant create the rules and tailor the expert system to the specific needs of the firm or individual. These packages provide an inference engine, user interface and explanation facility while the accountant provides the knowledge base. Many commercial packages of this type are available. Novices may use most of these packages with minimal training. This approach may be the most feasible, especially for smaller firms, because of its low cost and ease of use.
Expert Systems Application
In An Introduction to Artificial Intelligence and Expert Systems: A Management Advisory Services Special Report (AICPA), several accounting areas are identified in which applications of expert systems are possible. They are: audit planning, internal control analysis, account attribute analysis, quality review, accounting decisions, tax planning, management consulting and training. Exhibit 2 reviews the areas identified by the AICPA along with additional areas of financial statement preparation, governmental accounting and accounting education. Exhibit 2 also provides definitions and current applications that relate to each area.
Applications in Exhibit 2 entail the custom development of an entire expert system internally by an accounting firm or developer. These developed expert systems are not commercially available to the public for reasons mentioned in the previous section.
For any application area listed in Exhibit 2, an accountant can purchase a commercially available shell and add the knowledge base. With this approach, the accountant can use the shell to develop an expert system that addresses a specific need.
After reviewing various expert TABULAR DATA OMITTED system applications, accountants who have not seriously considered expert systems may realize that these systems have many applications in accounting. The applications examined are a clear indicator that expert systems have real potential to change the way accountants practice.
Expert System Benefits
Implementation of expert systems can benefit the accountant and the firm. The reliability of the expert system ensures consistent and accurate decision making. Specifically, some benefits of expert systems are:
* Expert systems make more consistent decisions than human experts. Expert systems do not have bad days, problems at home or feel pressure from the boss as human expert accountants do. These factors can influence a human expert's decisions, thus creating a higher variance in results.
* Expert systems are more reliable than human accountants. They do not get sick, take vacations or coffee breaks. Even the most stoic accountant enjoys a day off or becomes ill.
* Expert systems can reduce the number of staff. They can literally take the place of many accountants at a lower rate of pay, especially at middle-level positions.
* Expert systems increase productivity while decreasing time necessary to train. Expert systems increase productivity by allowing less experienced staff to take on greater amounts of more complex work. The staff is guided through these tasks by the expert system. Accountants can learn from this knowledge while performing assigned tasks.
* Expert systems preserve scarce knowledge. An expert system can capture the knowledge of a single individual who specializes in a certain area. This information can be utilized when the accountant is unavailable for a specific task due to other work, business trips, illness or retirement.
A Look to the Year 2000
If one looks ahead to the year 2000, one may not recognize the accounting firm as it is today. Potentially, expert systems will change the way these practitioners conduct business in a noticeable way. Three areas of change are: staff composition, basis of competition and the entry of new competition.
First, expert systems will change staff composition by reducing demand for senior and manager level accountants. Because expert systems capture knowledge from these accountants, lower-level accountants will perform some of the work previously performed by managers and senior accountants.
The second area of change is basis of competition. Currently, accounting firms compete based on integrity. Name and reputation are the most important factors considered when choosing a firm. Expert systems will change all this. In the future, the firm with the best technology could drastically reduce service expenses and undercut the competition by charging significantly less for services. Expert systems have the potential to reduce some accounting areas into a commodity-like service. This will hurt the smaller firms that cannot make the capital expenditures necessary to compete with the Big 6 internally-developed expert systems. The small firms must investigate various alternatives such as rule-based expert system packages, otherwise they may lose their niche in the accounting service market.
The third area of change concerns the entry of new competition. Currently accounting firms have a monopoly on accounting services. This may not be the case by year 2000, since new competition may enter the accounting service market by purchasing expert system technology. Although barriers to entry such as the integrity, professional requirements and professional standards exist, the potential for new competition is real.
Expert systems have not received the attention they deserve. In the past, many accountants were reluctant to accept computer and microprocessing technology. Today, many accountants are still reluctant to accept expert system technology.
By reviewing Exhibit 2 and observing the areas of application, particularly the current applications, even the harshest critics must concede that expert systems have already made an impact in accounting firms. Upon learning what an expert system is, what it can do and the benefits it can provide, accountants must realize that expert systems have the potential to change the operations of accounting practices. The firms that make the commitment now may be the firms on top in the year 2000. It is essential that accountants and their firms keep abreast of expert system technology since the time for expert systems is fast approaching.
Exhibit 1: Expert System Components
The components of the expert system are a knowledge base, an inference engine program, a user interface program and an explanation facility.
Knowledge Base: The area of the expert system that captures and stores the knowledge of human experts and other information sources.
Inference Engine: Computer programs that do the "thinking" for the expert system. The inference engine takes stored knowledge and processes information based upon the cues it is given.
User Interface: Provides communication between the system and end user. The keyboard and computer screen act as an interface to ask questions and deliver facts.
Explanation Facility: Explains to end users how answers are derived. The explanation facility is important for expert system success because having explanations assures end users of results and gives a feeling of comfort to the end user.
L. Craig Foltin, CPA, is the elected auditor for the city of Lorain, Ohio. He is a teaching fellow at Cleveland State University and a former assistant auditor for the state of Ohio.
|Printer friendly Cite/link Email Feedback|
|Author:||L. Craig Foltin|
|Publication:||The National Public Accountant|
|Date:||Jul 1, 1994|
|Previous Article:||FASB No. 116.|
|Next Article:||Accounting for subsidiary stock transactions.|