StarDust

Philosophical Fundamentals

As the StarDust project is first being formulated, there is enormous interest not only academic but commercial in “artificial intelligence” as Deep Learning effected by simulations of neural nets. Not for the first time, the idea that AI can best be achieved by study of and replication of the methods used in the human brain, has come to preponderate, to the extent now that the more formal approaches, often based in or related to formal logical notations, are scarcely mentioned, save to underline what researchers and application developers are not doing.

I believe this to be unfortunate. Most of the ways in which we have used computers are very different to the ways in which the human brain works, and there is every reason to think, not only that this will continue into the future, but also that methods quite dissimilar to neural nets (even Von Neumann Architectures) can make substantive contributions to the effectiveness with which we address the most difficult intellectual problems.

Be that as it may, it is not my purpose here, or that of the StarDust projects to advance that thesis.

Even if we seek only to re-implement and scale up human-like cognition, the capabilities which are addressed by StarDust are worthwhile applications of that kind of capability, and the possibility that addressing these applications in advance of the realisation of artificial general intelligence (AGI) might accelerate achievement of AGI may be thought of as a bonus extra.

The Development of Language and The Representation of Knowledge

Current AI orthodoxy is focussed on learning from sensory data to realise “knowledge’ represented as synaptic weights in simulations or simplified models of neural nets. The growth of higher knowledge depends upon the development of more sophisticated kinds of language, which are suitable for sharing knowledge, collaboration on its development and exploitation, and its application to reasoning about the environment and our interventions into the environment.

Communication has co-evolved with brain capacity. In the most primitive organisms we have communication by chemical signals, which serves to coordinate the behavioural adaptation of bacteria to their environment. I

Three Kinds of Proposition

The Deductive Paradigm Shift