Can computers reason？ Reasoning requires the individual to take a given set of facts and d

raw correct conclusions. Unfortunately, errors frequently occur, and we are not talking about simple carelessness as occurs when two numbers are incorrectly added, nor do we mean errors resulting from simple forgetfulness. Rather, we have in mind errors of a logic nature--those resulting from faulty reasoning. Now, or at least soon, computers will be capable of error-free logical reasoning in a variety of areas. The key to avoiding errors is to use a computer program that relies on the last two decades' research in the field of automated theorem proving. AURA (Automated Reasoning Assistant) is the program that best exemplifies this use of the computer.

AURA solves a problem by drawing conclusions from a given set of facts about the problem. Program does not learn, nor is it self-analytical, but it reaches logical conclusions flawlessly. It used various types of reasoning and, more important, has access to very powerful and sophisticated logical strategies. AURA seldom relies on brute force to find solutions. Instead it solves almost all problems by using sophisticated techniques to find a contradiction. One generally starts with a set of assumptions and adds a statement that the goal is unreachable. For example, if the problem is to test a safety system that automatically shuts down a nuclear reactor when instruments indicate a problem, AURA is told that the system will not shut the reactor down under those circumstances. If AURA finds a contradiction between the statement and the system's design assumptions, then this aspect of the reactor's design has been proved satisfactory. This strategy, known as the set of support strategy, lets AURA concentrate on the problem at hand and avoid the many fruitless steps required to explore the entire theory underlying the problem. Almost never does the program proceed by carrying out an exhaustive search.

The chief use for AURA at this time is for electronic circuit design validation, but a number of other uses will arise. For example, there already exist "expert systems" that include a component for reasoning. An expert system is a special purpose program designed to automate reasoning in a specific area such as medical diagnosis. These expert programs, unlike human experts, do not die. Such systems continue to improve and have an indefinite life span. Moreover, they can be replicated for pennies. A human who can expertly predict where to drill for oil is in great demand. A program that can predict equally well would be invaluable and could be duplicated any number of times.

Will the computer replace the human being？ Certainly not. It seems likely that computer programs will reproduce--that is, design more clever computer programs and more efficient, more useful components. Reasoning programs will also analyze their own progress, learn from their attempts to solve a problem, and redirect their attack on a problem. Such programs will assist, rather than replace, humans. Their impact will be felt in design, manufacturing, law, medicine, and other areas. Reasoning assistants will enable human minds to turn to deeper and far more complex ideas. These ideas will be partially formulated and then checked for reasoning flaws by a reasoning program. Many errors will be avoided.

The primary purpose of AURA is to

A．function as a safety mechanism in nuclear reactors.

B．detect contradictions and other faults in computer programs.

C．develop expert human programs for technical fields.

D．check human reasoning for possible errors.