Open-Ended Intelligence
A model of intelligence proposed by David Weinbaum and Viktoras Veitas (2015), as an alternative to narrow AI’s reward-maximization model and to humanist conceptions of intelligence as a fixed divine endowment. In Open-Ended Intelligence (OEI), “a distributed population of interacting heterogeneous agents achieves progressively higher levels of coordination” — not toward a predetermined goal but through “local resolution of disparities by means of reciprocal determination that brings forth new individuals in the form of integrated groups of agents (assemblages) that exchange meaningful information and spontaneously differentiate from their surrounding milieu.”
The key move is dismissing the “a priori givens” of intelligence: the agent, the environment, the distinction between them. Instead, identity is understood as metastable and provisional — “stations in an incessant process of transformation.” Since “the set of [any entity’s] interactive capacities is open and inexhaustible,” intelligence is an enacted process of openness and experimentation with the very faculty and limits of self. This constitutes an embodied pluralism — not relativism but practical pluralism under the purview of engineers and scientists.
OEI is used as a critique of the Moloch frame and coordination maximalism in web3: it argues that the web3 space needs procedures capable of self-reflexively interrogating the ontological assumptions animating its terms. The great Moloch — the force behind multipolar traps — isn’t merely coordination failure but the pre-rational ontological violence that constitutes agents as self-sovereign individuals in the first place. OEI offers a design horizon in which the reconstitution of the concept of agency is a technē — a creative, composable architecture open to distribution, disruption, and performance.