The Parity Test
A future of personal capability is only fair if participation does not depend on already arriving with the right stack.
01 / the institutional question
Access is not the same as standing.
Capability parity is the condition in which a person can participate fully whether they arrive with a mature stack, a modest toolset, or no personal AI capability at all.
The app-to-stack transition promises a genuine expansion of human range. A Stack Person can gather useful context, rehearse a hard conversation, compare sources, draft an alternative, or keep a private method improving across time. That is a better prospect than spending every working hour adapting to fixed fields and prescribed screens. But the promise can curdle if institutions quietly make a privately assembled stack the admission ticket for good work, good education, or civic participation.
App People were usually measured by their ability to operate the system chosen for them. Stack People will be measured, more often, by the quality of the questions they can frame, the judgment they can carry, and the capability they can assemble around a real task. Those changes are welcome. The problem begins when a school, employer, or public service assumes everyone already has the same contextual support behind the screen.
The parity test asks a simple question: if a person does not bring the preferred AI capability, can they still understand the task, receive the needed support, make a decision, and earn the same standing?
This is not an argument to ban personal stacks. It is an argument against confusing a useful private advantage with a legitimate condition of participation. A strong institution should make room for capable people to bring responsible methods while ensuring that someone without a stack is not converted into a second-class participant.
02 / run the test
Three arrivals. One standard.
Parity does not require identical workflows. It requires that the institution preserve a real route to understanding, contribution, review, and recourse for each person who has a claim to be there.
mature stack
Bring a method.
A person may arrive with carefully developed prompts, notes, routines, and models of their own work. The institution can set boundaries around records, security, and accountability without treating all personal capability as suspicious or forcing it to disappear at the door.
bounded contribution
basic tools
Use the provided route.
A person may use a managed assistant or a simpler set of approved tools. They should still be able to see the task, inspect the relevant information, ask for help, and make a meaningful contribution rather than merely accept an automated result.
supported contribution
no private AI
Remain fully present.
A person may choose not to use AI, lack access, or need an alternative for reasons of skill, privacy, cost, disability, policy, or conscience. The work cannot become unintelligible or the decision unreachable simply because their capability is different.
equal standing
03 / design for parity
Build alternatives into the work, not around the person.
Many institutions will reach for a crude answer: either require one official assistant or prohibit every personal tool. Both choices are easy to administer because both reduce variation. Neither asks the more valuable question: what capability must a person actually have in order to participate responsibly here?
That question leads to design rather than permission theater. A course can make its sources and evaluation criteria available even when some students use private study stacks. A workplace can preserve a reviewable system of record while allowing a worker to prepare privately from lawful, non-confidential material. A public agency can offer a human-readable path through an automated process, including a way to challenge an outcome. These are not identical settings, and they require different safeguards. The parity principle is portable even when the policy is not.
The argument extends The First Stack Generation: if capability becomes personal and cumulative, institutions must decide whether they will recognize the person behind the stack or only the people who can afford to arrive with one.
Can the task be understood?
Do not make a hidden assistant the only translator of the assignment, policy, or process. The underlying purpose, relevant information, and success condition must remain legible to the people asked to act.
Can the result be challenged?
A person needs a route to question an output, correct context, and receive a human explanation when the stakes call for one. Faster preparation is not a substitute for recourse.
Can contribution travel both ways?
Let a Stack Person bring a responsible method without demanding their whole private history. Let another person contribute without that method and still have their work taken seriously. Boundaries should protect relationships, not create a caste system of interfaces.
04 / the hard part
Parity is not sameness.
Capability parity does not mean every person gets every tool, every permission, or every result by the same route. A surgeon, a payroll administrator, a student, and a resident filing a public form have different risks, records, duties, and legal constraints. Some contexts should tightly control what enters and leaves. Some decisions should remain human-led. Some claims require documented evidence. None of that disappears because personal AI capability becomes more common.
Nor does parity turn individual preparation into an entitlement to bypass shared process. A Stack Person cannot use a private capability to move protected information into an unsafe place, evade review, or demand that everyone else accept an uninspectable answer. Personal capability earns institutional trust through clear boundaries, not through mystique.
But the symmetrical warning matters too. An institution cannot declare itself neutral while making participation depend on an assistant it owns, a workflow it will not explain, or a speed it only some people can purchase. If the process has become so automated that a person without the preferred stack cannot understand, contest, or complete it, the institution has not merely adopted a tool. It has changed who can stand inside the work.