Made-Up Minds: A Constructivist Approach to Artificial Intelligence

Two of the main goals of artificial in1 lenged this traditional approach. One of it, then we have a vicious circle between telligence understanding human cogthese is constructivism, which has spread the original and the knowledge of it. Innition, and building computer systems particularly in the last 20 years but whose stead, Piaget’s view is that “to know is to that perform as well as people in tasks sources can be traced to Kant, Vito, construct.” requiring intelligence can be mutually Berkeley, Montaingne, Phyrro, and even But to construct what? In knowing, he supportive. For example, an engineer the Pre-Socratics. says, we construct actions and intellectuwho wants to enable a computer to peral (mental) operations that he considers form tasks requiring intelligence might to be systems of transformations. These be aided by a viable model of human inare not copies of the transformations that telligence, while a researcher who wants constitute reality, he says, but consistent to develop a viable model of human inJean Picrget documents some and coherent models that are compatible telligence might verify and improve that striking uniformities with reality. Within these models, it is model by experimenting with an artificial throughout cogtdtive our experience that enables us to make I version. development, and refers to choices.7 Understanding human cognition these as the funttionul So, according to Piaget, the knowledge acquired by an intelligent organism is Especially in cognitive science and AI invarrirrnts of intelligence. compatible with the world (it organizes research, two requisites have been fundaDresrker proposes a schema the world) and is acquired by constructmental in all established theories of mwhanism that corresponds ing coherent systems of mental operaknowing:’ to Piaget’s theory of stages. tions and physical actions (it organizes l Whatever we would like to call “true itself).