AI+and+Expert+Systems

=2.4.2 AI and Expert Systems= toc

**Social and Ethical Issues**

 * responsibility for the performance of an expert system—knowledge engineer, informant, programmer, company that sold it, the buyer/consumer
 * value of the development of AI as a field, for example, whether it is an appropriate place to put economic resources
 * ethical issues of various applications of AI, for example, replacement of human workers, handing decision-making tasks to a computer
 * social impact of the use of “smart” machines on everyday life
 * ethical issues related to military applications of AI, for example, smart weapons, reconnaissance, decision making
 * implications of creative production by computers using AI, for example, Aaron, an expert system, creates visual art
 * access to the knowledge base underlying an inference engine in an expert system, for example whether people affected by decisions made using an expert system should have access to the rules by which the decision was made.

**Knowledge of Technology**

 * key terms—AI, Turing test, parallel processing, machine learning, natural language, common-sense knowledge, agent, pattern recognition, expert system, knowledge base, inference engine, heuristics, fuzzy logic, knowledge engineer, domain
 * storage requirements for common-sense knowledge
 * processing requirements for AI
 * collection/creation of a knowledge base
 * creation of an inference engine (for example, if/then rules, fuzzy logic)
 * identifying domains that are suitable for expert systems.

General Notes
//(taken from Computer Ethics)//

The **knowledge base** in an expert system is a structured understanding of acknowledged experts in a particular problem domain. For most part, expert systems are collections of rules that have been extracted from an expert by an **knowledge engineer**. The rules often (but not always) take the form of IF....THEN statements. For example, in a case of medical diagnostic system (MYCIN),:

>//IF the infection requiring therapy is meningitis,// >>//AND the type of infection is fungal// >>//AND the organisms were not seen on the strain of culture// >>//AND.......// >//THEN there is suggestive evidence that the Cryptococcus is not one of the organisms that might be causing the infection.//

IF...THEN rules are not the only for in which knowledge is stored in an expert system’s knowledge base. There are other forms however the essential nature of expert systems is applying DEDUCTIVE methods.

Expert systems not only use a **knowledge base** and an **inference engine** (to operate on the knowledge) but usually provide an explanatory **user interface** that justifies the conclusions of the systems line of reasoning with relevant probabilities.

The **knowledge engineer** has to be very skilled in identifying rules when gathering information from the expert. Also experts’ rules and knowledge may conflict, not only across experts but also even in the same expert. Resolving such clashes is also a part of a knowledge engineer’s task.

Storage requirements for common-sense knowledge
Common sense is the knowledge that can be used without any knowledge on a topic and are what our instincts tell us, things that are obvious. For example, the sky is blue. How do we store common-sense? One cannot just begin to read a book and achieve a lot of common-sense. Everyone has different levels of common-sense. Common-sense requires experience, and our memory stores it from our emotions or reactions to situations. You cannot exactly have a lot of common-sense without any facts. For example, you remember the time you were once burnt by fire, but it is not necessary to know the facts of the fire as one can tell that it is hot and needs to be careful with it.

Knowledge base
The knowledge base is the part of the expert system that contains all the facts and information that it needs to give solutions and solve problems. The knowledge base contains a system of rules for determining and changing the relationship among those facts. Some rules are straight-forward, for example, if salary is great than $50,000 then add income tax of 15% whereas some rules are **//heuristic//** (rule of thumb), for example if 'red meat for dinner, then select red wine'. The facts are stored in a **database** which are then organised in categories. An expert system’s knowledge base must be constructed by a user, an expert, or a knowledge engineer.

//__There are two types of knowledge bases:__//
 * Machine-readable knowledge bases store knowledge in a computer and they usually have automated deductive reasoning applied to them. They contain a set of data, which often comes in rules that describe the knowledge in a logically consistent manner.
 * Human-readable knowledge bases are designed to allow people to retrieve and use the knowledge they contain, primarily for training purposes. The main benefit from this knowledge base is that it can help a user find an existing solution to his or her current problem. An example of this may be the Microsoft online Help System.

Inference engine
An inference engine is a computer program that tries to derive answers from a knowledge base. It is the "brain" that expert systems use to reason about the information in the knowledge base for the ultimate purpose of formulating new conclusions. Inference engines are considered to be a special case of reasoning engines, which can use more general methods of reasoning.

An inference engine selects the next rule to apply & ensures consistency in that the same outcome would be derived if same information input.

In simple rule-based systems, there are two kinds of inference, //forward chaining// and //backward chaining//.

//Forward chaining// is when the data gets put into working memory. This triggers rules whose conditions match the new data. These rules then perform their actions. The actions may add new data to memory, thus triggering more rules. And so on. This is also called //data-directed// inference, because inference is triggered by the arrival of new data in working memory.

//Backward chaining:// the system needs to know the value of a piece of data. It searches for rules whose conclusions mention this data. Before it can use the rules, it must test their conditions. This may entail discovering the value of more pieces of data, and so on. This is also called //goal-directed// inference, or //hypothesis driven//, because inferences are not performed until the system is made to prove a particular goal (i.e. a question).

=Social and Ethical Issues=

Responsibility Scenario
//(adapted from Computer Ethics)//

A Nevada woman underwent routine surgery in a hospital. The operation was completed without complication. However soon after, she was administered pain relief by a computerised dispensing machine (a medical expert system). Unfortunately, the system mistakenly instructed hospital staff to pump more than 500 milligrams of pain-relieving drugs into her body and within 30 minutes of the successful completion of the operation, she went into a coma. Five days later she was dead. The patient’s lawyer launched a damages suit against the hospital for incorrect and irresponsible use of a medical system. Is the hospital responsible?

Who should pay?
Examine the roles of the stakeholders and for each, give reasons for/against as whether that stakeholder holds any responsbility for the issue: · the [|knowledge engineer] · the informant (expert/s) · the programmer · the company that sold it · the hospital who purchased the system