The Moving Baseline
When AI makes competent output cheap, the work standard does not stay still. The consequential advantage is the ability to raise, inspect, and redesign the standard itself.
Productivity can change the meaning of enough.
A polished first draft, an organized research brief, or a clean comparison used to signal unusual effort. As these become easier to produce, they can become the unremarkable starting point for the next request.
That is not proof that human work has become worthless. It is a description of a familiar economic motion: once a capability spreads, an institution quietly builds it into the floor. The client asks for the analysis and the alternatives. The manager expects the summary and the questions it provokes. The teacher sees a fluent answer and asks whether the student can defend the judgment behind it.
Capability inflation occurs when a new tool raises not only what people can produce, but what other people come to expect before they consider the work useful.
The phrase matters because it corrects a tempting story about AI. The story says an individual receives a faster tool, produces more, and therefore gets ahead. Sometimes that happens. But a tool widely available to everyone can turn yesterday’s advantage into today’s admission ticket. The visible output remains necessary; it is no longer sufficient.
App People experience this shift as a change in the assigned interface. A new feature appears. A workflow is revised. A dashboard adds another field. The organization raises the target and the person adapts inside the application. Stack People face a different question: which parts of their method should evolve so that a higher target produces better judgment instead of merely faster compliance?
The standard needs a human author.
If the baseline moves automatically, AI can become a ratchet: more output, less room to question its purpose. Stack People need an active relation to the standard—not just a faster route to it.
- Mark 01
Define the promise.
Say what the work is for before optimizing its format. Faster reports do not help if nobody has named the decision, person, or condition the report is meant to improve.
- Mark 02
Test the gain.
Ask whether the added speed changed quality, reduced a real burden, or only created a larger queue. A method becomes capability when it can notice that difference.
- Mark 03
Keep a limit.
Reserve places where the old baseline remains enough: a careful conversation, a checked source, a beginner’s learning curve, or time to reconsider a direction.
The scarce thing is not a faster answer.
When a team treats every improvement as a demand for more volume, the stack becomes an extraction device. The person learns the wrong lesson: conceal the method, race the clock, and avoid the questions that might slow the next delivery. That is an App People response to a system whose terms they did not help set.
A Stack Person can make a different move. They can retain a working standard: what counts as evidence, which audience is affected, which errors are costly, what requires a human decision, and what should be stopped rather than accelerated. This is not a proprietary prompt or a private archive of every thought. It is a revisable way of deciding what a good result means.
That ability compounds. A person who can explain why a baseline moved can decide whether to accept it, reshape it, or refuse it. They can bring a learned method into a new role without carrying confidential context from the old one. They can show a collaborator the relevant boundary without handing over their entire personal capability.
Organizations have a parallel responsibility. They should not treat a worker’s growing stack as an invisible subsidy that can be appropriated by raising targets indefinitely. The better question is not, “How much more can this person deliver now?” It is, “What new responsibility, learning, and agency should accompany this higher expectation?”