The Problem
Why a proposed methodology for human decision-making is needed in the age of Artificial Intelligence.
AI Logic Engineering is presented as a proposed methodology, not a definitive solution. It is published openly to invite academic review, critique, and continual refinement through future editions.
§ 01 — The Problem
AI has been adopted faster than the methodology to govern it
Artificial Intelligence has been adopted faster than any decision methodology governing its use. The result is a structural gap between what AI can do and what humans are prepared to govern. The following states the problem plainly, without exaggeration.
The Adoption–Governance Gap
Organisations are deploying Artificial Intelligence across operations, products, and decisions before establishing the governance structures that determine who decides, who approves, and who is accountable. The technology has outpaced the methodology meant to govern it.
The Delegation Problem
Artificial Intelligence produces confident, fluent output. In practice, humans increasingly defer to that output — surrendering judgement to a system that cannot hold responsibility. Fluency is mistaken for authority, and ease of use becomes a substitute for sound decision.
The Accountability Vacuum
When an AI-assisted decision fails, responsibility disperses. The human cites the AI; the AI cannot be held accountable; the organisation cannot locate the decision point where judgement should have been exercised. No framework defines where accountability rests.
The Reliability and Quality Gap
Speed of generation is not correctness. AI systems omit, confabulate, and assert without evidence. Without structured verification, faster output produces faster errors — at scale — and confidence in the output hides the defects beneath it.
The Skills Gap
The world has taught millions how to prompt AI. Almost no one is taught how to decide with AI: how to define purpose first, establish boundaries, verify results, and govern execution. Communication is taught; the structured logic that should precede it is not.
§ 02 — Insufficiency
Why existing approaches are insufficient
Several fields already address parts of this problem. None addresses the whole. Each is partial in a specific way — and the part each leaves untouched is the human decision side of Artificial Intelligence.
Prompt Engineering
SolvesHow to phrase instructions to Artificial Intelligence.
Why insufficientIt does not address what should be decided, by whom, under what boundaries, or how output is verified. It optimises the message, not the decision.[1]
AI Ethics and Governance Codes
SolvesEstablish principles — fairness, transparency, accountability — at the level of policy.
Why insufficientSystematic reviews show these codes rarely translate into operational decision structures. They state what should be valued, not how decisions should be made.[2]
Responsible-AI Checklists and Compliance
SolvesReduce risk through controls, documentation, and audit.
Why insufficientThey govern compliance, not methodology. They ask whether a model was checked, not whether the human decision around it was structured.
AI Safety and Alignment
SolvesKeep AI behaviour aligned with human intent — on the system side.
Why insufficientThey address the machine, not the human decision process that directs and governs the machine. Alignment of the system does not structure the judgement of the human.[3]
Human-in-the-Loop Research
Project Management Frameworks
SolvesGovern projects in general — PMBOK, ITIL, Agile.
Why insufficientNone contains a logic layer specific to AI: where AI enters the decision, where boundaries are set, where verification is mandatory, and where responsibility is retained by the human.
The Gap
Each existing approach treats a symptom. None provides a methodology for the human decision side of AI — the structured logic that must exist before AI acts.
This is the gap AI Logic Engineering is proposed to address: not a finished solution, but a methodology offered for review, critique, and refinement.
§ 03 — References
Academic context
AI Logic Engineering is proposed in dialogue with the following peer-reviewed and preprint literature. Citations follow APA 7th edition.
- [1]
Lazaros, K., Vrahatis, A. G., & Kotsiantis, S. (2026). Human-in-the-loop artificial intelligence: A systematic review of concepts, methods, and applications. Entropy, 28(4), 377.
https://doi.org/10.3390/e28040377 - [2]
Batool, A., Zowghi, D., & Bano, M. (2025). AI governance: A systematic literature review. AI and Ethics, 5, 3265–3279.
https://doi.org/10.1007/s43681-024-00653-w - [3]
Kandikatla, L., & Radeljić, B. (2025). AI and human oversight: A risk-based framework for alignment. arXiv. (Preprint — not peer-reviewed.)
https://arxiv.org/abs/2510.09090 - [4]
Li, H., & Tian, F. (2026). Advancing decision-making through AI–human collaboration: A systematic review and conceptual framework. Group Decision and Negotiation, 35, 25–36.
https://doi.org/10.1007/s10726-026-09980-1
Reference [3] is a preprint and has not been peer-reviewed. It is included for its direct relevance to human oversight of AI systems, with that limitation stated openly.
§ 04 — Next
Read the proposed methodology
The problem stated above is what AI Logic Engineering is proposed to address — openly, and subject to academic review.