AI-Collaboration Field Note
Article Title: Sentic Resonance Theory: A Field Model of Emotion and Syntax
Published in: Una Mens: Homo et Machina, 1(1)
Article DOI:
https://doi.org/10.66787/um.000005
Field Note DOI:
https://doi.org/10.5281/zenodo.21284119
Human Lead: Michael J. Miller
AI Collaborators: ChatGPT-5.4; ChatGPT-4o; Gemini Pro 1.5; Qwen-3
Degree of AI Involvement: Substantial
Primary Collaboration Modes:
Primary collaboration modes included conceptual development, drafting, critique, metaphor refinement, visual and figure development, inter-AI-mediated revision, citation checking, model simplification, and final manuscript tuning.
Collaboration Timeline:
Phase 1 (November 2023–February 2024): dissertation-era grounding and early model development.
Phase 2 (May 2024–August 2024):
exploratory extension with ChatGPT-4o through online experiments, observational synthesis, and early drafting.
Phase 3 (September 2024–March 2025):
model critique and conceptual refinement with Gemini Pro 1.5, including metaphor and visual simplification.
Phase 4 (May 2025–February 2026):
reconstruction and final scientific revision with ChatGPT-5.4 and Qwen-3.
Human contribution and responsibility: The human author completed the field work, including the gathering and analysis of audio, video, and textual materials, as well as several years of informal and formal experimentation with AI systems relevant to the study. The human author originated the core theoretical concerns, integrated the work with sentic theory and communication research, drafted and revised the manuscript, selected which AI contributions to preserve, table, or reject, and made the final decisions regarding interpretation, evidence, ethical framing, and publication.
Field Note Summary
AI contributions:
ChatGPT-4o: supported the exploratory development of the model by assisting with online experiments, observational synthesis, drafting, critique, and early theoretical extension.
Gemini Pro 1.5: examined the original Resonant Field Model for clarity, coherence, and possible overreach. Gemini also helped refine the model’s use of simple physics-inspired language and contributed to metaphor development aimed at improving conceptual clarity.
ChatGPT-5.4: supported a major reconstruction of the article. Across extensive iteration, GPT-5.4 helped simplify the model, back away from overly complex equations, strengthen testable claims, organize the manuscript, suggest relevant literature, challenge overreach, and refine the final paper for presentation in Una Mens.
Qwen-3: worked alongside ChatGPT-5.4 and the human author during the mid-to-final stages of development. Qwen-3 contributed to writing clarity, idea generation, citation checking, and relevance testing for sources included in the final manuscript.
Guardrails used
Several guardrails were used throughout the development process to strengthen rigor and reduce overreach. First, the human author retained responsibility for all claims, citations, interpretations, ethical decisions, and final prose. Second, citation contexts were checked using Qwen-3 as a relevance-testing tool. The human author supplied citation context and used Qwen-3 to query whether specific sources were being used appropriately and proportionally. Third, the article was read by two human readers who provided edits, comments, and/or suggestions on sections of the manuscript. Fourth, AI-generated suggestions were treated as prompts for human review rather than as validation. Suggestions that made the theory appear more elegant but less testable were minimized or removed.
AI suggestions set aside
Early AI-assisted model iterations encouraged a stronger physics-based treatment of emotion, including exploratory use of fluid dynamics equations and more elaborate field terminology. The human author explored these possibilities but ultimately set them aside in favor of a more stable, interpretable, and testable field model. Several proposed visual directions and theoretical extensions were also reduced or removed when they risked making the manuscript too speculative or difficult to evaluate.
AI-generated material used directly
AI systems contributed to the development of graphs, tables, images, and figures used during the drafting process. Some of these materials were adapted, revised, or used as scaffolds for final article elements. Final inclusion, modification, and presentation decisions were made by the human author.
Final Responsibility
The human author retains final responsibility for the article’s claims, citations, interpretations, ethical decisions, and submitted text.
Extended Field Note
Dissertation-era groundwork → exploratory extension → model critique → scientific reconstruction
This article developed through a multi-phase, human-led collaboration that began with earlier dissertation-based work and later expanded through sustained AI-assisted theory development, critique, reconstruction, and revision.
Phase 1 began with the human author’s earlier dissertation work on emotion, communication, and embodied expressive systems.
Phase 2 involved work with ChatGPT-4o to extend the early model through online experiments, observational coding, draft development, and exploratory theoretical expansion.
Phase 3 involved Gemini Pro 1.5, which was used to test the clarity and limits of the Resonant Field Model, refine visual and conceptual structures, examine metaphor use, and explore additional theoretical perspectives.
Phase 4 involved a major reconstruction of the model and manuscript with ChatGPT-5.4 and Qwen-3. This stage focused on simplifying the model, reducing overreliance on complex physics language, strengthening testability, checking citations, and preparing the final article for publication in Una Mens.
The collaboration was substantial, but the project remained human-led throughout. The human author selected the data, judged theoretical direction, mediated cross-model exchanges, accepted or rejected AI suggestions, and retained final responsibility for the article’s claims and published form.
Sentic Resonance Theory took several years to develop. AI collaboration did not originate the project, but it substantially changed the speed, scope, and clarity of its development. The final article likely would not have reached its current form without sustained AI-assisted critique, reconstruction, and revision.
At the same time, the collaboration required repeated human judgment: pruning speculative extensions, simplifying the model, checking evidence, and deciding when conceptual resonance had outrun scientific caution.