中文Your team is faster than it was a year ago. The decks are cleaner. The emails are sharper. The analysis arrives in seconds instead of days. And something else is happening underneath all of it, quietly, in a place no dashboard measures: people are starting to outsource the thinking itself.
Researchers call this cognitive offloading, and when it becomes a reflex rather than a choice, it creates what Dr. Michelle Rozen — “the Change Doctor” — names as cognitive debt. You get speed now, and you pay for it later in atrophied judgement, diminished ownership and a workforce that has quietly lost confidence in its own mind.
The public conversation about AI and work is stuck on the wrong question: will the machine take the job? The more immediate one is what happens to a human who stops practising. Thinking is a capability, and capabilities behave like muscles — built through resistance, decay without it. Every time a professional wrestles with an ambiguous problem, sits in the discomfort of not knowing, forms a point of view and gets it wrong, that person gets stronger. Remove the struggle and the reps disappear — and nobody notices they are gone.
Cognitive debt shows up in small, deniable moments. The blank stare in a meeting when a simple follow-up question reveals the output was generated but never understood. The sentence “that’s what the AI suggested” — the sound of accountability leaving the building. Homogenised thinking, as everyone queries the same models and regresses toward the mean. And, most damaging, erosion of confidence: people stop offering the half-formed idea or the instinct built over fifteen years because the machine sounds more certain than they feel.
The loss of confidence is the real casualty. Judgement is not just reaching the right answer; it is the willingness to stand behind an answer under pressure, with incomplete information and no tool to tell you what to do. That willingness is built on accumulated experience. When people stop doing hard things, that foundation cracks — and a workforce without conviction is a queue of people waiting for instructions.
Most leaders run one of two failed playbooks. The first is ban it: restrict the tools, police usage, pretend it is containable. That produces shadow usage and dishonest employees, not protected thinking. The second is unleash it: hand out licences, celebrate productivity, declare victory. That is the more expensive mistake, because the gains are immediate and visible while the costs are delayed and invisible. You book the efficiency this quarter; you pay the debt in a year, in a currency you did not budget for.
The answer is deliberate AI — deciding, on purpose, which thinking your people must keep doing themselves, and protecting it like the asset it is. Rozen offers five moves. First, draw the line between leverage (using AI to compress work around the thinking) and abdication (using it to replace the thinking). Second, require the point of view first: form your own answer before you ask the machine, then use AI to pressure-test it. Third, ask the second question — interrogate reasoning, not output. Fourth, protect the struggle for junior talent who most need the reps. Fifth, make ownership non-negotiable: every piece of work has a human name on it, and “the AI wrote that” is not a defence.
AI is the most powerful thinking tool ever built, and it will make your organisation faster whether you manage it well or not. The question is what kind of humans you have on the other side of that speed — people who bring a point of view to the machine, or people who wait for it to give them one. For print-business owners wrestling with where to deploy automation across prepress, estimating and customer service, the lesson lands hard: protect the judgement that differentiates your shop, and let the machine handle the drudgery around it. Use the tool. Keep the muscle. The shops that thrive in the next decade will be those that treat AI as a junior analyst to be managed, not a boss to be obeyed.
Source: American Printer (Dr. Michelle Rozen, “The Change Doctor”), published 14 July 2026.
For print-business owners, the framework translates into concrete shop-floor habits. A estimator who lets AI generate a quote without ever sanity-checking the assumptions about substrate waste or labour time is building a fragile business; a prepress operator who accepts AI-generated colour corrections without understanding why they were made will struggle when the model faces a job outside its training distribution. Rozen’s “point of view first” rule is especially relevant where margin lives or dies on judgement: form the estimate, the schedule or the spec yourself, then use the tool to stress-test it.
The risk is not hypothetical for smaller printers, who lack the layered review processes of large organisations and often wear multiple hats. When the owner is also the estimator, the designer and the account manager, the temptation to offload thinking is highest precisely where accountability is most concentrated. The antidote is cultural, not technical: naming explicitly which decisions remain human, rewarding people who catch the machine’s errors, and reserving genuinely difficult problems for junior staff to wrestle with. The shops that thrive in the next decade will be those that treat AI as a junior analyst to be managed, not a boss to be obeyed — amplifying the judgement that differentiates them rather than quietly surrendering it.
