AbstractPhilosophy has recently witnessed a radical critique of the epistemological tradition centered on the notion of representation. It is here argued that such a critique is relevant to the current endeavors of Artificial Intelligence, both in its broad (i.e. as an attempt to elucidate human cognitive capacities) and narrow (i.e. as a practical concern to develop ‘intelligent’ systems, e.g. expert systems) definitions. It is shown that many of the recurrent problems faced by AI researchers stem from their espousal of the old paradigm conceptions of ‘knowledge’, ‘understanding’, ‘justification’, and the like. Some suggestions concerning a possible alternative for such conce