Tactical Analysis, Communication Theory

I. Summary

The shared question of the three papers published in 2024 is how the rapidly growing volume of data can be converted into shared knowledge that genuinely supports the adaptation of a military organisation and responsible human decision-making. The summary is based on GRIN's public abstracts, tables of contents and previews [1–3].

Shared theoretical core

In Dombrádi's approach, warfare is not merely the simultaneous use of tools and branches of arms, but a communication system: a temporally linked chain of sensing, information selection, interpretation, coordination, action and feedback. Decisive advantage is therefore created not by data superiority in itself, but by how quickly the organisation can arrange signals from different sources into a shared, actionable knowledge pattern. The systems-theoretical background highlights the importance of internal self-observation, observation of the environment, and recurring communication.

The communication theory of combined arms warfare [1]. The study describes the operation of combined arms as a coordination and knowledge-transfer problem. Different capabilities form a genuine system only if knowledge organised at a higher level is interpretable locally, tasks are synchronised in time, and feedback is fed back into the next action. Accordingly, the goal of training is not merely the drilling of procedures but the formation of shared knowledge patterns, communicative routines, and adaptive cooperation.

The role of artificial intelligence in military decision-making [2]. In the "information hurricane," AI can compare patterns quickly, consistently and in great volume, but the authors distinguish between this linear pattern recognition and the human's knowledge patterns, built from socialisation, experience, meaning-making and self-reflection. AI's strongest role is therefore in decision preparation: uncovering correlations, gaps, contradictions, temporal dependencies and alternatives. Interpreting context, weighing goals and values, and responsibility remain with the human decision-maker.

Natural and artificial neural networks, and the training of the air force [3]. The third paper starts from the differing strengths of the two kinds of learning. The artificial system is scalable, repeatable and effective on large datasets; human learning, however, is embodied, social, meaning-sensitive, and adapts more flexibly to unexpected situations. The direction for renewing training is therefore the AI-based analysis of data from simulations, war games and exercises, and then feeding the results back into theory, training situations, and the next experimental cycle.

Summary thesis: the basic unit of modern military decision support is not the standalone algorithm, but the shared learning cycle of the human, the organisation and AI: data → synthesis → meaning → decision → action → feedback.

II. New research plan

Adaptive decision-preparation and data-synthesis system – human-centred AI for processing multi-source, uncertain and contradictory military information.

Goal: to create an auditable research prototype that produces, from heterogeneous data, not an automatic command but a decision-preparation picture equipped with sources, uncertainty, and alternative explanations.

Research question and hypothesis

To what extent does multi-source, source-critical AI data synthesis that displays uncertainty reduce decision-preparation time while preserving or improving accuracy, verifiability and human situational understanding? The hypothesis is that the machine is stronger in large-scale comparison, the human in context and responsible choice; their structured cooperation yields better results than either alone.

The five modules of the prototype

• Data and provenance management: unifying open, synthetic or authorised data; preserving timestamp, source, quality and uncertainty.
• Semantic synthesis: organising events, actors and dependencies into a knowledge graph; separating mutually reinforcing and contradictory statements.
• Hypothesis and alternative generation: producing multiple interpretations and decision variants, with assumptions, consequences and uncertainty bands – without automatic execution.
• Human–AI collaboration interface: explainable summaries, traceable sources, information-gap indicators, and visualisation to support the decision-maker's follow-up questions.
• Practice–learning cycle: analysis of logs from simulations and war games; feeding errors, surprises and successful patterns back into the next training and model-development cycle.

Method and evaluation

Scenario-based experiments should compare groups performing the same task with traditional versus AI-supported work. Metrics: decision-preparation time, recall of relevant information, recognition of errors and contradictions, calibration of certainty, mental workload, trust, source use, and group-level resilience. The tests should also include incomplete, delayed and misleading data.

International connections and outcome

The plan connects to NATO STO's directions in AI, big data, visualisation and decision support [4], NATO's federated data-sharing strategy [5], and the JWC's human–AI wargaming experiments [6]. DARPA ITM emphasises alignment with human decision values [7], the Army Research Laboratory's HAT emphasises task allocation and learning across missions [8], Dstl emphasises data engineering, human–machine cooperation and assurance [9], and RAND emphasises the combined use of modelling, simulation and wargaming [10]. NATO ACT's 2026 direction preserves human decision authority alongside faster data synthesis [11].

Expected outcome: a closed, auditable demonstrator; a source- and uncertainty-aware data model; a synthetic test set; a human–AI experimental protocol; training recommendations. The system may not make an independent operational or weapons-employment decision: final judgment and responsibility remain in every case with a designated human decision-maker.

III. Source list

Krisztián Dombrádi's military-science works

• Dombradi, K. (2024). The Communication Theory of Combined Arms Warfare. Munich: GRIN Verlag. 13 p. Catalog No. V1488108. ISBN 978-3-389-04519-0. MTMT record
• Varga, A. & Dombrádi, K. (2024). The Role of AI in Decision Making for Military Operations. Munich: GRIN Verlag. 21 p. Catalog No. V1495581. ISBN 978-3-389-05378-2. MTMT record
• Dombrádi, K. (2024). Transdisciplionarity Practical Comparison of Natural and Artificial Neural Networks. Thoughts on the Air Force Renewing Training System. Munich: GRIN Verlag. 11 p. Catalog No. V1498397. ISBN 978-3-389-06117-6. MTMT record

Related international research directions

• NATO Science and Technology Organization (2025). Collaborative Programme of Work 2025. Information Systems Technology Panel: AI, big data, visualization, decision support and human–machine teaming.
• NATO (2025). Data Strategy for the Alliance. Federated data spaces, metadata, data-centric security and data-driven decision-making.
• NATO Joint Warfare Centre (2025). From Strategy to Code: How AI Is Reshaping Military Wargaming. Human–AI wargaming experiment and structured evaluation of LLM support.
• Defense Advanced Research Projects Agency. In the Moment (ITM). Human-aligned AI decision-making in high-stakes, ambiguous domains.
• DEVCOM Army Research Laboratory. Human Autonomy Teaming (HAT). Resource allocation, human-guided machine learning, team assessment and mission-to-mission learning.
• Defence Science and Technology Laboratory (updated 2024). AI and Data Science: Defence Science and Technology Capability. Data engineering, analytics, visualization, human–machine teaming and assurance.
• Davis, P. K. & Bracken, P. (2022). Artificial Intelligence for Wargaming and Modeling. Journal of Defense Modeling and Simulation. doi.org/10.1177/15485129211073126
• NATO Allied Command Transformation (2026). Next Generation Targeting Project. Federated AI and data management for faster synthesis with human judgment retained.

Source note: the summary of Dombrádi's works is based on GRIN's freely available bibliographic data, abstracts, tables of contents and public previews.