For years we associated competence with production: writing a report, preparing slides or analysing a spreadsheet. AI has made part of that production fast and abundant. This does not mean everyone has become more competent. Packaging has become cheaper. Content, judgement and accountability still have a price.
Well-formatted mediocrity
A generative system can produce formally correct prose in seconds. If the question is poor, the result is often elegant mediocrity: readable, tidy and devoid of a real position. The risk is not only error but convergence. When everyone asks similar systems similar questions, language becomes smooth and ideas lose friction.
Uniqueness comes from context
What distinguishes a person is not access to a tool but the depth of context they bring: experience, failure, observation, relationships and responsibility. An experienced professional sees what is missing. They know which data should not be generalised and which organisational consequence is absent from the slide. AI can amplify that competence; it cannot retrospectively invent its history.
Asking is work
A good question bounds the problem, makes constraints explicit and defines the level of evidence required. Above all, it accepts that the desired answer may be wrong. In healthcare this discipline is essential: a convincing prompt is not validation, and a fluent response is not a source.
Responsibility cannot be automated
If a text, decision or project carries our name, responsibility remains ours. Saying “AI wrote it” is the contemporary equivalent of blaming a calculator for an incorrect balance sheet. Leadership in the algorithmic age requires less worship of tools and more capacity to verify, choose and accept consequences.
Becoming more human, not slower
The answer is not to reject AI. It is to use it to release cognitive time and invest that time where machines are weaker: listening, negotiation, situated experience, moral responsibility and trust. AI does not make you unique. It can make what is already distinctive about you more visible.
Build a recognisable voice
Using AI well requires editorial discipline. Experience is gathered first; then a thesis is formed, sources capable of contradicting it are sought and only then is the machine assigned work it can accelerate. The text must preserve the author’s vocabulary, doubts and responsibility. If removing the name makes the piece indistinguishable from thousands of others, the problem is not model style. It is the absence of a position.
The advantage of memory
A personal archive of cases, decisions, failures and sources is more valuable than an infinite library of prompts. Memory recognises what returns under new names and connects present enthusiasm with consequences already observed. This is where experience becomes an advantage: not in knowing everything, but in knowing which questions the present tends to forget.
Technology does not replace responsibility. It makes it harder to hide.
This content is educational and is not clinical guidance. Editorial reconstructions are distinguished from documentary material.