A theoretical goal of artificial intelligence (AI) is to reproduce in some fashion the thinking process of the human mind. I say “in some fashion” because the actual processes of the human brain are only partially known at present. Research on AI started in the 1950s, but has been--and continues to be--constrained by the limitations of computer hardware, both processing speed and storage capacities. Growth in both enabled increasingly powerful systems to be built; even so, today’s systems remain relatively narrowly focused on specific areas. No system yet covers the broad scope of knowledge and reasoning capability that each of us carries in our head.
There are various ways to approach creating a general-purpose AI system, one that has similarities to what the human mind can do. Research on biological brains reveals information about the neurons that comprise the brain. Software that simulates the processing done by a network of neurons, called a neural network, produces systems that can be trained on sets of data. Such training permits the neural network to analyze and recognize similar data, thus producing a system that can, for example, perform various types of medical diagnoses. Although such systems can in some cases produce results more accurately than human experts, there is no consideration that these systems think in the way a human does. But, there is interest in producing a thinking system.
The virtual mind is a short book that presents a model and design for such a thinking system. The model is based on psychological theory and research about thinking from antiquity to the 21st century, chiefly beginning with the work of Freud and Jung but primarily using the theories of Matte Blanco. As opposed to neural nets or the work of Ray Kurzweil (see his 2012 book How to create a mind ), which consider brain processing from a deep functional perspective, The virtual mind considers what the brain does rather than how it does it.
There is merit to both approaches; ultimately they must merge to some extent to create an entity that can approximate human thinking. The psychological approach examines the mind’s perceptions and its relation to the human community, the social aspect providing context for absorbing and reacting to what is perceived, including emotional reactions. Hence, the model accumulates a database built from public sources to establish a zeitgeist, or “public spirit” against which inputs can be tested and provided context.
Following the development of the psychological background for a design, The virtual mind presents the design and functional components of a computer-based system to implement the model. A data model, processing components, and data flows are described, with diagrams showing how they connect internally. The book concludes with a brief discussion of why to go this route, a very broad and general suggestion of application areas, and some of the challenges that remain, such as that mind functions are only partially understood, meaning that any current model will necessarily be incomplete. One question that comes to mind is what practical applications might be realized from this implementation. The author says there will be a sequel describing application to robotics.
One can question the reason for mimicking a human mind, especially the emotional component. Given human history, it is reasonable to ask if we are the best model to use. The book does not address this. Further, it is unclear that implementing a model of the human mind as conceived by extant psychological theories will produce a thinking intelligence given our current state of knowledge. Modern research has shown quite clearly that the mind consists of a conscious part and an unconscious (I prefer calling it “subconscious” because it is active though not directly perceived) part, the latter clearly active and in communication with the consciousness. It is the “unconscious” that remains mysterious.
The virtual mind is an interesting, if brief, exploration of the history of psychological theories of mind and how an AI based on these theories might be implemented on a digital computer system using off-the-shelf components such as Java and MS SQL Server. The book should be of interest to anyone curious about or involved in AI. I also recommend to those interested in this area Kurzweil’s book noted above, a much more detailed exploration of the functional aspects of a brain’s operation that presents theories Kurzweil has employed in his widely used inventions.