AI-Assisted Knowledge Creation Systems Using Generative Models

Abstract

AI-assisted knowledge creation -- the use of generative AI systems to augment, accelerate, and enrich human knowledge production across scientific, professional, and educational domains -- represents a paradigm shift in the relationship between artificial intelligence and human cognition. While much attention has been devoted to generative AI as a content generator that replaces human writing, the more strategically significant application is as a knowledge co-creation partner that amplifies human expert capacity by handling knowledge synthesis, literature integration, structural organisation, and knowledge gap identification tasks. This paper proposes the Generative Knowledge Creation Assistant (GKCA) framework, a system architecture for AI-assisted knowledge creation that integrates five specialised generative modules: a scientific literature synthesiser (SLS) that generates structured summaries of large document collections; a knowledge gap analyser (KGA) that identifies unexplored research questions from corpus analysis; a hypothesis generator (HG) that proposes novel research hypotheses grounded in existing knowledge; a concept relationship mapper (CRM) that constructs and visualises knowledge graphs from generated syntheses; and a knowledge validation assistant (KVA) that checks generated knowledge claims against scientific literature. GKCA is evaluated through a longitudinal knowledge creation study involving 64 domain experts (researchers, knowledge managers, and educators) who used GKCA across three domains -- biomedical research, technology policy, and environmental science -- over 90 days. GKCA users produce 3.2x more knowledge artefacts per unit time than control users working without AI assistance (p < 0.001, Cohen d = 4.18), with expert-rated quality scores of 4.2/5.0 (SD = 0.4) for GKCA-assisted outputs versus 3.8/5.0 (SD = 0.6) for unassisted outputs (p < 0.01). The study contributes the GKCA specification, a Knowledge Creation Quality Index (KCQI), and empirical evidence for the substantial productivity and quality benefits of AI-assisted knowledge creation in expert domains.