Applications: where Preservative AI operates

The same inversion runs through every domain: where traditional AI deployment substitutes, accelerates, and smooths, the preservative deployment anchors, audits, and surfaces. Six domains, each with the contrast made concrete.

The contrast at a glance

Journalism

The traditional deployment treats the article as raw material: summarise it, headline it, repackage it for feeds — every step a substitution that moves readers further from sources, and a framing-amplification besides. The preservative deployment treats the article as evidence about its own construction: the audit links the original prominently (Principle 1), analyses what the headline does to the issue, runs the Marginalised Voices lens over the sourcing (who is quoted; who is spoken about), and publishes the omission map beside the piece. The reader gets the article plus its silences — never a substitute for it.

Policy analysis

"Extract key points and predict impacts" accepts the policy document's own framing as the unit of analysis — which is precisely what a policy document is engineered to make you do. The preservative analysis starts behind the document: the Deep History lens maps which decisions and failures produced this proposal (and which alternatives were discarded before the public ever saw a draft); Suffixscape flags where the drafting language diffuses responsibility ("it was determined", "implementation will proceed"); the Future Modelling lens asks whose futures the impact projections centre and who was consulted. The deliverable is the policy made contestable: its history, its authors, and its absences restored.

Education

Study-guide generation is epistemic flattening with a syllabus: the primary text — with its difficulty, ambiguity, and voice — is replaced by a smooth derivative, and students learn the derivative. The preservative deployment inverts the relationship: the AI's output is a reading companion that anchors to the primary text, marks where the guide simplifies ("this summary collapses a contested debate — see the exchange itself"), and applies the Artistic Perception lens to preserve what the text's form does that no summary can. The deeper application is pedagogical: students taught to run framing audits and Suffixscape passes themselves are being taught to audit framing rather than consume content — the literacy the Adoption Guide centres for educators.

Research

Literature-review synthesis inherits and amplifies the citation network's existing exclusions: the already-cited get re-cited, and the synthesis presents the resulting consensus as the field. The preservative deployment makes the network's shape an explicit object of study: citation exclusion tracking (which work, which regions, which languages, which methodologies never enter the loop), methodological blindspot mapping via the Scientific Evidence lens (what the dominant methods structurally cannot see), and synthesis outputs that carry their gaps section as prominently as their findings. The review's claim changes from "here is what the field knows" to "here is what this corpus, with these exclusions, supports".

Corporate communications

The traditional tooling drafts and polishes — it is, candidly, an agency-diffusion machine, optimising tone precisely by removing friction and accountability from sentences. The preservative deployment points the instruments the other way: Suffixscape audits the release for nominalised evasion and missing agents before publication; the Marginalised Voices lens tracks which stakeholders the communication systematically omits. Used internally, this is an integrity tool (say who decided, attach the receipts to the superlatives); used externally — by journalists and advocates on corporate output — it is an accountability tool. Both uses are intended.

Community knowledge

Translate-categorise-index is where flattening does its deepest damage: oral, relational, and contextual knowledge forms are converted into the database's categories, and what didn't fit the schema simply ceases to exist in the searchable record. The preservative deployment treats the form as part of the knowledge: preserve oral and contextual structures in their own shape, flag every point where a Western categorical scheme replaced a community's own organisation (Cross-Cultural Wisdom lens), and put governance where the CARE Principles put it — with the community, including the right not to be indexed at all. Here more than anywhere, the methodology's own limitations apply: the practice supports community authority; it does not substitute for it.

The common pattern

Across all six domains the deployment question is identical: does the system replace the source material or anchor to it; does it amplify the framing or audit it; does it smooth the absences or surface them? Any domain not listed here — law, medicine, archives, local government — can be derived by asking those three questions. That is what it means for Preservative AI to be a methodological layer rather than a product.