There is a newer version of the record available.

Published February 18, 2026 | Version v1
Publication Open

KIS: A Question-Centric Protocol Architecture for Hierarchical AI Thought Control

Authors/Creators

Description

We propose the Knowledge Innovation System (KIS), a five-tier hierarchical architecture for AI thought control that shifts the paradigm from answer-centric optimization to question-centric protocol design. The five tiers are: (1) Command, (2) Context, (3) Agent, (4) SuperAgent (question-space exploration with protocol-as-subject), and (5) Autonomous SuperAgent (self-revising protocol). We formalize tier distinctions using category theory (functors and natural transformations), lattice theory (join/meet on question spaces), and Jacobian eigenvalue analysis for dynamic cognitive mode selection. Adversarial validation across five AI systems (GPT-4o, Gemini, Perplexity, Claude) confirmed logical coherence, and a literature review found no equivalent unified protocol in the 2024-2025 research landscape. KIS is implemented as a 200,000+ character protocol with nine cognitive modes, pseudo-parallel processing, and a semantic fermentation model. We discuss implications for AGI cognitive architecture and propose an external computation engine for future implementation.

Keywords: AI thought control, question-centric design, meta-agent, category theory, lattice theory, protocol-as-subject, cognitive architecture, semantic fermentation, SuperAgent

Files

kis_paper.v2.pdf

Files (1.3 MB)

Name Size Download all
md5:c23d379973067642420451998e0db7cd
1.3 MB Preview Download

Additional details

Additional titles

Translated title (Japanese)
KIS: 問い中心パラダイムに基づくAI思考制御の階層的プロトコル設計

Dates

Issued
2026-02-18

References

  • Wei, J., et al. (2022). Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. NeurIPS.
  • Yao, S., et al. (2023). Tree of Thoughts: Deliberate Problem Solving with Large Language Models. NeurIPS.
  • Madaan, A., et al. (2023). Self-Refine: Iterative Refinement with Self-Feedback. NeurIPS.
  • Significant Gravitas. (2023). AutoGPT: An Autonomous GPT-4 Experiment. GitHub. Varela, F.J., Thompson, E., Rosch, E. (1991). The Embodied Mind: Cognitive Science and Human Experience. MIT Press.
  • Hutchins, E. (1995). Cognition in the Wild. MIT Press.
  • Mac Lane, S. (1998). Categories for the Working Mathematician. 2nd ed. Springer.
  • Davey, B.A., Priestley, H.A. (2002). Introduction to Lattices and Order. Cambridge University Press.
  • Zhang, Q., et al. (2024). Agentic Meta-Orchestrator: Task-Adaptive Multi-Agent Orchestration. arXiv.
  • Wang, L., et al. (2024). A Survey on Large Language Model Based Autonomous Agents. Frontiers of Computer Science.