How can AI systems deal with large and complex problems?

Problems arising in the near future may grow larger and more complex beyond the human capability for managing them and it is worrying that this can ca use trouble for human society. The only possible way to solve this problem is to make computers more intelligent to back up the weak points of human b eings. At least, it is necessary to provide a method for capturing and reco rding all human decisions in problem-solving, This view leads one to an ide a of a computer-led interactive system such that the computer becomes able to manage problem-serving process and makes records of individual decisions made by a human being. Autonomy, generality and practicality are the major requirements for such computer-led interactive systems. New information te chnology to meet this condition is discussed. It is very different from the conventional computer technology in many aspects. The major topics are: a model-based software architecture for future information systems, a modelin g scheme to accept and represent a wide area of problems, a method for exte rnalizing human ideas and of representing it as a model, a method for solvi ng problems represented in the form of a model and so on. It outlines a par t of the research work being promoted by the author's group under the spons orship of the Science and Technology Agency of the Japanese Government.