Svět z pohledu počítačů

The control of complex technical and economical systems depends on the quality of information which we have about these systems. Complex systems have mostly hundreds of sensors with different accuracy and reliability characteristics (see [2]). The measured data are mostly given with a precision lower than two decimal places (see [12],[13],[14]). At the present time problems connected with the control of complex systems, i.e. problems based on data corrupted with uncertaintes, are beyond the ef- fective reach of the control theory. The fundamental dilemma of this situation lies in the fact that the uncertainty cannot be simply included in formal theories. In the paper the failure of formal theories by the control of complex systems is discussed in the framework of the philosophical theory of the knowledge. Two concepts are compared with respect to their historical grounds. The first - the Leibniz concept of the knowledge - is based on matematics. The uncertainty of information cannot be considered. So formal theories based on this concept cannot be used in scientific branches which start from uncertain data (control science, information theory, artificial inteligence, system engineering). Here the second concept - the modified Kant concept of the knowledge - must be used because only this concept respects uncertainties in input data, i.e. the limited accuracy of measured data ([20],[21],[22]). So we have tried in this paper to show that including uncertainties in the science should lead to the new paradigma not only in technical sciencies but in matematics and philosophy too.