In conclusion, to compare the NBI technology with the state of the art of the field, we will briefly discuss the analogies that may be drawn between the NBI technology and the functioning of brain neurons as is currently known. It is no longer a revelation that construction of the so-called "artificial neural networks" is not much different from traditional programming and is steadfastly turning into a collection of connectionist models. The training of an artificial neural network consists in adjustment of weights assigned to neuron inputs. In doing so, the modern neurocomputer technology does not take pain in pondering over those countless "leftover nuts and bolts", i.e. details in other respects well known in modern science - morphological and functional peculiarities of somas, dendrites and axons, electrical signals in the form of sequences of narrow impulses of constant amplitude and varying frequency, as well as many others. The artificial neural network concept can be boiled to down to presenting a neuron as a basic "input - memory - output" element. In this context, it would be hard to anticipate that the evolution of artificial neural networks may proceed in a way even remotely similar to biological evolution.

The essence of the NBI technology is information processing performed through exposure of information to iterative transformation operations (synaptic impulses?). Completion of processing may require from tens to thousands of such iterative operations. If we theoretically assume that a process of this type may occur in one neuron, then dendrites of other neurons should be receiving the enriched, sorted and specialized information. Thus functioning a neuron would be able to substitute for myriads of "brainless" neurons of artificial neural networks.

The illustrations presented in this report cover some of the features of the software based on the non-biological intelligence (NBI) concept. It should also be noted that the NBI-algorithm is capable of recognizing various patterns and sequences, including texts, with hundreds of variables. It can search for and effeciently find not only identical but also - what is more challenging - analogous objects. The techniques, including those already developed by us and currently under development, employ numerous methodological tools, such as deriving equations to define relationships between clusters in a hierarchical system, selection of relevant variables, real-time construction of various types of dendrograms and trees, a selective search within large databases, and many others.