Generative AI as Support Infrastructure: A Design Theory of Inclusive Intelligence
Аннотация
Generative AI (GenAI) is increasingly deployed as support infrastructure, yet its capacity to accommodate heterogeneous and fluctuating user conditions through interactional design remains underdeveloped and unevenly realized. This study asks how GenAI can be designed to sustain usable, appropriately bounded support amid such variability. This question is addressed through an echeloned design science research (DSR) process organized into a four-stage theory-development pipeline, using career support for professionals with ADHD as a critical case in which inclusive design requirements become empirically tractable. The evidence base comprises one naturalistic online subcommunity corpus and four participant samples spanning controlled and ecological evaluation. Across four stages, we specify the problem space, derive empirically grounded design requirements, operationalize them as interaction policies within a configured large language model (LLM), and evaluate the resulting policies through controlled and situated evaluation. These stages culminate in a design theory of inclusive intelligence , instantiated through MATRIX, a set of six interaction-level design principles (momentary regulation, ability amplification, trust and stability, resonant attunement, intentional flourishing, and experiential integration) that govern how usable, contextual, and bounded support is enacted in practice. Inclusive intelligence is thus advanced as a mid-range design theory that specifies how GenAI support is calibrated and bounded as user conditions change, locating value in interactional governance rather than model capability. Its transferability is bounded to episodic, text-mediated support characterized by cognitive and affective variability, where guidance must be structured, paced, and delimited under shifting demand.