Knowledge, Skill and Artificial Intelligence

The term “information society” was first used in a 1972 Japanese futurological study (Japanese Computer Usage Development Institute 1972), in which the time perspective was limited to the year 2000. If the visions of this study were to become reality, a “computer mind” should have been established through education by the year 1985. A different perspective emerges from a new Japanese futurological study from 1985 (National Institute for Research Development 1985), which contains an example that illustrates the effects of a longer period of computer usage on one’s confidence in one’s actions.

Fundamental to the design of knowledge-based systems is the understanding of the nature of knowledge and the problems involved in computerizing it. This chapter deals with these issues and draws a distinction between three different categories of knowledge: propositional knowledge, skill or practical knowledge and knowledge of familiarity. In the present debate on “Information Society”, there is a clear tendency to overemphasize the theoretical knowledge at the expense of practical knowledge thereby completely ignoring the knowledge of familiarity. It is argued that different forms of theoretical knowledge are required for the design of current computer technology and the study of the practice of computer usage. The concept of dialogue and the concept of “To Follow a Rule” are therefore fundamental to the understanding of the practice of computer usage.

The nature of nursing has been the subject of discussion for the last 10–15 years. One reason is that in many countries the education of nurses has moved from teaching hospitals to academic institutions. This move has given rise to the question of the scientific basis for nursing knowledge.

Behind the concept of the automated factory lies the view of knowledge in which all human thought and action can be logically described in a formalized language, and in which all conceivable activities are predictable. So far all attempts to realize this vision have refuted this as being the case. The problem with automating propositional knowledge can be illustrated in analogy to the use of formulae at the expense of practice which means an enormous loss in association material when it comes to tackling problems. It is important to limit use of formalizations to ensure freedom of action through the ability to have a feel for things. Present production techniques do not provide an environment for the development of skills in training hence the retention of an apprenticeship model is fundamental as “The experiences of the senses and practical intellect are perhaps both an essential complement and prerequisite of the successful automation of industry”. (Emile, Rousseau)

Increased information through computerization does not mean greater reliability in decision making. Consider the case of weather forecasting. Skill here is the ability to select and interpret information. Computer solutions are general and standardized whereas skill is developed through use of concrete, specific examples. Traditionally, meteorologists used solely historical material to build up a “comprehensive idea” of the weather situation in their minds. This “comprehensive” idea is the shaping of an inner weather picture which gradually builds up in the minds of meteorologists and leads to understanding and development of skill. Its creation depends on the reflection and digestion of information. Forecasting is a continuous process. The problem with the new methods and information presented by computerization is that there is less importance placed on reflection and therefore understanding through consideration of a variety of knowledge sources, communication between workers, and practical experiences.

What is “Tacit knowledge”? What role does it play in working life? What implications does it hold for understanding, for example, implications of expert systems for working life? What do philosophers of science have to learn from working life studies?

When real practical problems need to be solved by people in a particular social and economic context, the technology has no more than an instrumental function. Technology itself is politically, socially, and economically neutral but is political in its use, reflecting and possibly strengthening the value system and the economic and political interests of those who control it. This has implications for concepts entailed in the idea of transferring skills through its use. The preconditions for the transfer of skills are the motivation of the recipient, shared practical context, some common language between those imparting and receiving skills. There are severe limitations as to how far the computer can satisfy these conditions. The issue of the transfer of skills should be of central importance in any political programme. For what future are people to be trained? What skills are to be transferred, and to whom? Who is involved in preparing a solution to perhaps the central problem of our era and who is left to form part of the problem?

One of the central planks of development is the enhancement of people’s potential for creative participation in the development process by enhancing the provision of basic human needs such as education, health and welfare. If AI technology has any relevance to these developmental issues (as like any other technology it should), then it must also concern itself with the knowledge of human needs and the nature of expert knowledge which contributes to the design of relevant technologies. In order for people to participate in the development process, they must have knowledge and skills to affect such a process. The machine-centred approach of current AI technology restricts its application to those problem-solving domains which can be formalized in logical rational rules. This approach thus takes account of non-intuitive knowledge and ignores the intuitive component which is embedded in the personal experience and in the social and cultural traditions of the user. The human-centred approach, on the other hand, is based on the human-machine symbiosis and provides for creative participation of users in the design of “developmental” systems.

Skills are necessarily representative of value systems of which they are a part. Hence the transfer of skill has to be considered as the transfer of a skill and part of a value system without which the intrinsic skill is meaningless. Knowledge transfer is better termed knowledge exchange. At present, training (intrinsic skill acquisition) is being divorced from education as a whole (location of skills in a value system) leading to the danger that skilled groups will lose touch with the social value system within which its skills are located, and probable erosion of skill. The intelligent system can be another powerful tool of the expert or it can be designed to be transparent to the non-professional. To ensure the democratization of knowledge the system must be transparent i.e., the learner is at the centre of this process such that acquisition and personalization of a skill is achieved by transforming what one observes another doing into something of value to oneself.

Important issues need raising before detailed knowledge acquisition can take place for designing expert systems. It is futile to try acquiring knowledge and constructing systems unless it is known what it is for.

At present there is a huge growth of medical knowledge with increasing specialization in specific fields. This poses a great problem for the general medical practitioner (GP), the sole generalist in medicine, who is unlikely to be able to handle this growth in ten years’ time if it continues at this rate. Public expectations and criticism of doctors have increased as people become better informed through the various types of media. Hence there is pressure for the use of computers in the medical profession to assist the doctor, thereby increasing his/her competence.

We are now at a unique historical turning point. Decisions we make in respect of the new technologies will have a profound effect upon the way we relate to each other, to our work, and to nature itself. Vast computer systems, expert systems, and artificial intelligence systems should not be seen as a technological bolt from the blue. They are in fact part of the historical continuum which is discernible in Europe certainly over the last five hundred years. Scientific and technological change, viewed historically, does seem to embody three predominant historical tendencies. Firstly, there is a change in the organic composition of capital. We tend to render processes capital-intensive rather than labour-intensive. Secondly, it constitutes a shift from the analogical to the digital. The manner in which we perceive our world, analyse it and relate to it is dramatically changed. Thirdly, it is a process in which human beings are rendered passive and the machines become more active. We recall “the more you give to the machines the less there is left of yourself”. It is against this historical background that there is an urgent need to view alternative systems, in particular those which may be regarded as human-centred. This paper will describe such human-centred systems.

The aim of this chapter is to consider the limits to rationalization and, therefore, the barriers to further devaluation of professional skills. If these limits are not recognized, an erosion in special qualifications acquired through professional training may emerge, thereby adversely affecting the quality of services. It is important to look at how far-reaching are the consquences of using information technologies as a means to distribute tasks and functions of administrative work and to substitute human labour for the devaluation of skills. Examination of two case-studies leaves little doubt that increased rationalization through automation of various parts of the decision-making process causes erosion of skill and professional knowledge, increased division of labour destroying that element of skill gained from the organizational culture, and hence limits the scope for action (inflexibility as technology does not accept the contradictions of social reality) and the ability to act with competence in a context. Only through “careful” strategies of rationalization can flexibility and effectiveness of administrative action be maintained or improved in an increasingly complex environment.

This chapter discusses the changing nature of apprenticeship training and describes some of the principles and weaknesses of the system of supplying engineering craftsmen in the UK. The “new” approach to craft training in the UK focusses on aspects such as duration, contents, standards, assessment, relevance, and openness. For example, training needs are increasingly seen in terms of mending and mental rather than making and manual skills and knowledge. This development requires that job structures should transcend existing rigid job boundaries and take into account such features as versatility, adaptability, and diagnostic ability. While there is a move to create a high degree of common skilling amongst engineering craftsmen there is also the recognition of a need to produce both “skill” and “process system” specialists. New apprenticeship in the UK is being developed with three recognizable streams: specialist craftsmen; operator craftsmen; and, process systems craftsmen.

This chapter deals with issues such as those of delegation of decison making, quality of service and competence, arising from the application of information technology in the domain of the Norwegian National Insurance Administration system. For example, within the context of “national conformity” of practice, the quality of service will demand uniformity in interpretation regardless of the individual insured person’s particular circumstances. The emphasis on efficiency and time, and a degree of “tailor-made” information may lead to reduced access and communication between the people who want to use the system and the service itself. Rationalization in complex domains (cases) of decision making may cause the case-handler either simply to ignore the problematic aspects of a case, or to make an intuitive choice instead of conferring with the relevant legal sources. There is empirical evidence from the Norwegian case that in a large number of cases the use of data systems may have a positive effect as they can weed out elementary errors and omissions in calculations, but at a more complex level where legal expertise and judgment is required, such systems may give no insight and support. It is important, therefore, to develop criteria for evaluating the non-time-related aspects of the quality of service, and thereby define aims for, and evaluate, public-oriented activities.

There is an abundance of promises to expert systems with user-friendly and intelligent interfaces providing cloned expertise for augmented job performance. A closer look reveals that very few systems are in operation, and that there is a serious gap between claims and reality. Basically, “knowledge engineering” is a modern form of “work study engineering”, and yet job and organizational design issues are rarely addressed in the literature on designing expert systems. This serious gap is discussed.