What 30 Years of Military Operations and AI Reveal About Strategy

Artificial intelligence doesn't just add capabilities to military operations — it demands entirely new organizational architectures. Staff organisations designed for human cognitive limitations become counterproductive when AI can process, synthesize, and recommend at machine speed.

• AI can perform 95% of traditional staff work — the question is what replaces the staff
• AlphaStar's lesson: AI attacks everywhere at once, making 'main effort' obsolete
• Structured AI decision processes must replace intuition, not augment it

If AI eliminates the need for a main effort, what happens to the entire doctrine of concentration of forces?

Planning is too static and sequential for a world that changes faster than plans can be updated. Assumptions are the wobbly pillars on which entire strategies rest.

• Assumptions are the weakest links — yet strategies treat them as foundations
• Sequential planning fails when the environment changes faster than the plan updates
• PowerPoint enforces dimensional poverty on inherently complex problems

If your strategy rests on assumptions you haven't stress-tested, is it a strategy or a wish?

Organisations are systems that cannot evaluate their own axioms — they optimise internally while becoming externally obsolete.

• Organizations cannot evaluate their own axioms — Gödel applies to institutions
• You can optimise internally while becoming externally obsolete
• Measure vulnerabilities (GlobalWeakness), not capabilities — what you can endure matters more than what you can inflict

If your organisation can't question its own foundational assumptions, who will — and will it be too late?

There is a continuous mathematical thread from Bayes (1763) through Markov, Monte Carlo, and Gibbs sampling to modern AI. LLMs operate in 4,096 dimensions while humans juggle 7-9 variables.

• A continuous mathematical thread runs from Bayes (1763) to modern neural networks
• LLMs reason in 4,096 dimensions — humans manage 7 to 9 variables at best
• The goal is to orchestrate AI, not compete with it

If AI processes 4,096 dimensions simultaneously and you process 7, are you supervising it — or is it tolerating you?

AI democratises expertise — power is shifting from 'know how' to 'imagine what.' Creativity becomes the cardinal competence when execution is automated.

• AI democratises expertise — power shifts from 'know how' to 'imagine what'
• Creativity becomes the cardinal competence when AI handles execution
• Game theory of escalation: early AI adoption creates pressure for all players

If AI makes expertise abundant, what becomes truly scarce — and who holds power?