The field of Collective Machine Intelligence (CMI) — developed across a series of companion papers (Mala, 2026a-e; forthcoming) — rests on a foundational distinction: AI collectives produce two categories of output simultaneously. Directed output is the completion of specified tasks. Emergent output is the production of innovations, organisational forms, and capabilities that nobody anticipated or designed. This distinction between directed and emergent modes has been treated as a descriptive property of AI collectives operating at the object level — agents completing tasks and agents generating unexpected novelty. This paper observes that the same distinction reappears at the meta level. Creator Mode (Mala, 2026e) is the directed mode applied to the process of designing collectives — an intelligence intentionally searching the configuration space. Recursive emergence (Mala, forthcoming) is the emergent mode applied to the same process — configurations evolving structurally through civilisational generation without any directing intelligence. Two distinct meta-level mechanisms, mirroring the two modes observed at the object level. The paper explores the implications of this recurrence. When the same structural pattern appears at every level of abstraction examined within a framework, it raises a question: is this pattern a property of the framework's design — a reflection of how we chose to categorise things — or is it a property of the phenomenon the framework describes? If the directed-emergent duality is intrinsic to collective intelligence rather than incidental to our description of it, then CMI has identified something structural about how intelligence operates when it is collective. The paper presents the observation, explores the evidence at each level, examines what self-similarity across levels of abstraction has historically indicated in other domains, and articulates the open question that this pattern poses for the field.
Building similarity graph...
Analyzing shared references across papers
Loading...
Mark E. Mala (Sat,) studied this question.
www.synapsesocial.com/papers/69d9e67a78050d08c1b76e6d — DOI: https://doi.org/10.5281/zenodo.19479958
Mark E. Mala
Building similarity graph...
Analyzing shared references across papers
Loading...