Do I understand correctly, that for each new task I should additionally train any system based on the UMNICK Digital Brain?
Not necessarily – it depends on the type of the new task. For example, if you have set the task of training the system to hammer nails into two boards, then you will need: 1. To show several different boards – to teach the concept of “board.” 2. To show several nails – to teach the concept of “nail.” 3. To show how to hammer nails into a board. Later, the Digital Brain’s abstracting function will allow the Digital Brain to hammer practically anything similar to a nail into anything similar to a board. But if in the future you set the task of screwing a screw into two boards, then the System will more than likely hammer the screws like nails – it does not yet know what a screw is and what it means to screw in. In this case it is necessary to additionally train – people also invented and began to correctly use screws much later than nails.
