Published in 1985 by Editura Tehnică, the book Artificial Intelligence appeared at a time when the field was only sparsely represented in Romanian specialist literature. Its purpose was to provide a systematic introduction to the ideas and methods of artificial intelligence as they had taken shape in the classical phase of symbolic AI: problem solving, symbolic representation, search strategies, and knowledge-based systems.
During the 1980s, artificial intelligence was consolidating internationally as a distinct field at the intersection of logic, mathematics, computer science, and the first cognitive models. In Romania, access to external sources was limited, and the publication of a book explicitly devoted to AI also had an educational role: to establish a vocabulary, clarify the scientific stakes of the field, and provide a coherent map of its themes and methods.
The book was written together with Cristian Giumale. Chapters 1, 4, 5, and 6 are mine, while Chapter 2 and part of Chapter 3 were written by Cristian Giumale. The final chapter was added at the suggestion of Ion Iliescu, then director of Editura Tehnică, in order to complete the work with a synthesis of the problem-solving methods used in artificial intelligence.
The conceptual thread of the book is explicit and rigorous at the same time: the intelligence of a system is not reducible to “results” or to the imitation of human behaviour, but is defined by the mechanisms through which a system can formulate, represent, and solve problems. From this perspective, artificial intelligence becomes an engineering discipline of symbolic reasoning: the construction of internal models of problems and the application of search and control strategies to those models.
The book emphasizes what remains, even today, a fundamental lesson of AI: the performance of a method depends critically on the chosen representation. A problem is viewed as a state space in which a solution is obtained by applying operators, or transformations, and by exploring that space under the control of a strategy. Thus the notions of state, operator, path, strategy, and evaluation criterion are not merely terms, but components of a problem-solving model.
In a realistic setting, many problems cannot be approached through general algorithms offering complete guarantees, and the book introduces the role of heuristics as mechanisms for guiding search: rules or criteria that reduce the explored space and make it possible to obtain a solution within a reasonable time. The importance of heuristics is discussed without rhetoric: they are not “magic,” but a form of encoded knowledge used to control the search process.
The classical fields of AI are presented as variations of the same conceptual core: problem solving, games, automated theorem proving, language processing, pattern recognition, expert systems, and intelligent robots. The book implicitly suggests the methodological unity of these applications: they differ through the type of representation and control they use, not through an abandonment of rigour.
Seen retrospectively, this work marks my entry into AI as a natural continuation of my interests in theoretical computer science and software design. The emphasis on models, representations, and control would later reappear in my reflections on computer-aided design systems, software process modelling, and, later, intelligent systems applied to healthcare.
