Automated Decision Making

In his article "Automated Decision Making", Davenport discusses the historical reluctance to adopt automated management decision making system in organizations. In the past, key decision makers did not feel comfortable letting computers make complex decisions for the organization. Their skepticism was mostly because of the complexity of decision making and the fact that real world decisions could not be reduced to a few simple variables. Another reason why managers resisted making these systems a part of their organizations was that they would be too complex for most users to understand. Even systems that were developed to support managers in the decision-making process were difficult to use because of the quantitative skills they required.
However, these systems are now becoming commonplace within many organizations for two reasons. First, advances in information technology have made it possible for these systems to solve more complex business problems. Secondly, business decisions have become increasingly more complex, to the point that many of them cannot be adequately analyzed by human beings. These new systems do not require much human involvement, which means that organizations do not have to assign specialists to manage them.
Automated decision-making systems are still only useful for solving certain types of problems. They are only practical for applications where there is a large supply of electronic information. These systems also can only solve problems that are well understood and the methodology for handling them is clear. Also, the standards of defining optimal decisions cannot be too subjective. Automated decision-making systems are also most appropriate for situations where it is important to make a decision quickly.
On the other hand, automated decision-making systems can be very useful in solving problems where there is a need for consistency. The likelihood that they will make a mistake is much lower than that of a human being solving the same problem. Human beings are also more prone to look at each problem on a case by case basis and use their own judgment when solving problems. This makes automated systems very useful when there is a need for consistency. They can also pick up on sensitive data and sense changes that would be easily overlooked by human employees.
Even though automated systems still require access to significant amounts of electronic information, it is being increasingly used for applications that are not highly quantitative. However, even though the application does not seem to be highly quantitative, it still relies on quantitative data. For example, Davenport discusses how a major winery uses sensors to monitor temperature and other weather conditions and passes this information to automated systems. These systems are then able to make certain decisions such as how much water to provide when irrigating the grapes in the vineyard. However, the more complex decisions still must be made by human operators.
Despite the fact that automated systems have provided accuracy, consistency and timely decisions for managers, they have also introduced other problems for organizations. The largest problem that executives face is that they must make sure that the standards, limits and variables for the systems that they are using. If they are not well understood, the system will still deliver a solution based on the information it was provided, but most likely it will not be the solution that executives are looking for. Davenport discussed how Cisco Systems made poor assumptions with the automated inventory system it was using to manage inventory, which ended up costing the company over $2 billion. In this situation, not only did the company poorly define the information that would be used in the system, it did not monitor the system sufficiently.
Managers must also recognize the importance for recognizing the need for exceptions. Since automated systems must operate consistently, humans must be able to oversee the systems and identify situations where the decision determined by the automated system must be overruled. If computers do not have enough data to make a reliable decision, they must be programmed to inform a human to handle the decision. Davenport states that some organizations, such as hospitals punish their employees for not overruling these systems. He argues that managers need to understand why these decisions were made. Another problem with these systems is the need to find experts who are capable of setting up and managing these systems. Davenport mentions an insurance company that had to discontinue using its system because it did not have anyone to manage it.
Legal and political issues can also affect how automated systems will be used in following years. Davenport mentions a hospital that was sued over a decision made by an automated system, and how future lawsuits may discourage companies from using these systems. Laws are also being made to not only regulate the use of automated systems but also the access that companies may have to personal information and how this information may be used by companies that make decisions that may affect the individuals.
Davenport’s main point in this article is that automated systems are becoming more widely used, and have some distinct advantages over using human beings to make certain decisions. However, he stresses that there are limitations in these systems, and has shown many companies have made tragic mistakes by not overlooking these limitations and hiring employees to monitor them appropriately. While there are many possible factors that can contribute to these problems, the biggest reason that automated systems often provide poor decisions is that the variables and situations which they are employed are not properly defined or set up. Therefore, companies must carefully oversee the automated systems that they are employing and make sure that the problems they solve are well understood.