How pressure narrows both what a person can do and the horizon through which people receive one another.
The project began by asking how people become less legible to themselves under pressure. It now follows the same narrowing outward: into conduct, institutional interpretation, records, data, artificial intelligence, and the conditions returned to the next person.
This changes the scale of the work. The narrowed horizon is not only an inner limit. It can become transmissible, procedural, durable and recursive.
Each stage can appear to be merely receiving what came before. The project makes the sequence visible so responsibility does not disappear between the handovers.
The person-level and system-level accounts are not competing explanations. They continually produce conditions for one another.
Conditions affect energy, regulation, safety, honesty, self-trust, meaning, responsibility and viable action. A person may know more than they can presently express, or protect capacity through conduct that looks different when its sequence is removed.
The question is not simply what the person did. It is what conditions shaped the range of actions, language and trust still available to them.
Workers, services, families, groups and institutions also operate through narrowed or widened horizons. Their capacity to tolerate uncertainty affects what they notice, how quickly they infer motive, what enters the record and whether correction remains possible.
The same person may be received very differently depending on the horizon through which the field interprets them.
Narrowing travels when an unfinished interpretation becomes conduct and the conditions produced by that conduct are later treated as evidence for the original view.
AI did not create the narrowed horizon. It entered the loop quickly enough to accelerate it, legitimise it and return it as apparently neutral interpretation.
The Route Before the Chatbot: AI-Supported Reflection, Polluted Care Language, and the Narrowed Human Horizon begins before the technology question. It asks what happened in the human route before a person opened a chatbot, what language fields AI inherits, and what conduct is required to keep judgement answerable.
The document develops accountable containment, closure without absolution, sequence restoration, polluted language for inner states, interpretive restraint, speaker-defined care, the right to correction, public-service stakes and a practical calibration standard.
Calendar age is the wrong measure. Consequence depends on how quickly a tool enters routes that already carry authority.
AI did not need decades of professional use to become consequential. It only needed to enter drafting, summarising, triage, assessment, moderation, education and public-service preparation while the language and record fields beneath those uses remained insufficiently examined.
The relevant combination is speed, scale, authority and recursive re-entry.
An AI-shaped interpretation can influence a record, decision or first understanding. That new language may later become data, organisational memory or training material. The system is not only inheriting yesterday's assumptions; it may help produce tomorrow's.
The question is therefore not whether a human clicked the final button. It is how the field was shaped before the human judgement arrived.
The project follows the point where useful language becomes overauthority: when a possible interpretation gains more confidence, reach or durability than the evidence can carry.
Words such as resistance, avoidance, defensiveness, insight, boundaries and dysregulation may be useful. They become polluted when detached from the sequence, evidence and relational conditions needed to use them accurately.
A response to a particular route becomes a stable property of the person. Future workers meet the inherited trait before they meet the living sequence.
A person must be able to see, challenge and correct language used about them without the correction itself becoming evidence against them.
Widening is not permissiveness, agreement or the removal of responsibility. It is the restoration of enough reality, sequence and capacity for responsibility to become viable and accurate.
Safety, regulation, language, route clarity and proportionate burden change what the person can honestly understand and do.
Observation, inference and missing context are kept distinct long enough for the person, conduct and surrounding field to become legible.
Conduct may need challenge, limits or consequence. The person is not reduced to a category, frozen inside past evidence or denied the possibility of newer conduct.
Judge conduct where necessary. Understand conditions. Compare current evidence. Preserve the person.
The architecture is shared across the site, but each route protects a different part of the sequence.
How adult wording carries or damages child signal when a child cannot yet answer the record back.
Development as becoming, and the need for language that does not convert unfinished life into fixed interpretation.
How interpretations become conditions through conduct, and how whole-field standards keep policy answerable to people.
What happens when lived knowledge is invited, recognised, used, constrained, withdrawn from or severed by hierarchy.
The public calibration route for using AI as a limited thinking chamber without surrendering judgement or human signal.
How knowledge is received, translated, circulated, applied and kept answerable to its source.
The project is a developing architecture rather than a single thesis page. These documents mark its current public foundations and live history.
AI-supported reflection, polluted care language, recursive narrowing and practical calibration standards.
The opening public account of self-legibility, pressure, adaptation and the organised narrowing of human range.
The public threshold document connecting capacity-first policy claims with the conduct people actually encounter.
A project-history note on recognition, widened possibility, ethical demand and institutional re-narrowing.
The Narrowed Human Horizon is not presented as a finished intervention brand or a total explanation of human conduct. It is a public architecture being developed through lived sequence, cross-field scrutiny, ethical restraint and the requirement that interpretation remain correctable.