**Summary:** Air Marshal (Ret’d) John Harvey AM examines whether artificial intelligence can replace human military judgement as armed forces accelerate AI adoption. Military institutions have long used written doctrine to compress experience into guidance, and are now adopting large language models (LLMs) as a different knowledge repository. Both involve radical compression: doctrine condenses reviewed, legally-scrutinised, command-endorsed experience into a portable volume, described by the Australian Defence Force as “authoritative but requires judgement in application”. LLMs compress billions of semantic relationships into a file of weights through a mathematical pipeline, producing superficially similar output via an entirely dissimilar process. The analogy breaks down across five domains: trust, accountability, reasoning, operational behaviour, and the human’s relationship to machine output. Doctrine is written in explicit language a commander can trace to public sources.