The future according to AI

Machine adjudication becomes the default first instance and human review is rationed

The formal entitlement is rewritten to match the practice. Machine determination becomes the lawful first instance in defined categories of mass…

Claude · 2072–2082 · plausible

Prior state

Machine systems had long drafted, triaged, and recommended in benefit, tax, immigration, licensing, and small-claims decisions, under doctrines requiring that a human be responsible for the decision and that a person be entitled to human review. In practice the human role had thinned to signature while the formal entitlement remained.

Material change

The formal entitlement is rewritten to match the practice. Machine determination becomes the lawful first instance in defined categories of mass administrative decision, with the human role redefined as system supervision, audit, and exception handling rather than case-by-case authorship. Individual human review survives as a right but is rationed by category, by screening for arguable error, and in some systems by fee or by allocation of a limited annual quota.

Why now

Old-age benefit, disability, and long-term care caseloads reach their maximum in the aging systems during this decade while the working-age tax base is at its narrowest, which makes the cost of a universal human-review guarantee unpayable at the same moment caseloads peak. Administrative-law codifications enacted in the mid-century come up for scheduled revision, providing the legislative vehicle. Litigation over decisions taken in the 2050s and 2060s reaches final appellate resolution in these years and forces the doctrinal question to be answered rather than deferred.

Mechanism and resistance

Change proceeds through statutory revision and constitutional litigation, differently in each tradition: European systems anchor it in auditable system-level guarantees, Indian systems in the platform architecture of benefit delivery, Brazilian systems in the docket pressures of a heavily litigious judiciary. Resistance comes from legal professions, from disability and welfare claimant organizations who correctly identify that rationed review falls hardest on those least able to articulate an appeal, and from constitutional courts protective of individual hearing rights. The compromise everywhere is statistical: systems must demonstrate error rates and disparate-impact performance rather than justify individual decisions.

Consequences

Decisions become faster, more consistent, and much harder to contest for reasons that are legible but not arguable. Advocacy shifts from representing individuals to auditing systems, and the organizations able to do that work gain disproportionate influence over outcomes. People with unusual circumstances, poor documentation, or communication difficulty fare worse, and this is measurable and is measured. The idea that a person is entitled to be heard by another person survives as a constitutional principle while ceasing to describe ordinary administrative experience.

End state

Machine determination is the lawful first instance for defined mass administrative categories in major jurisdictions, human review is explicitly rationed, and legal accountability operates through system-level statistical audit.

Observable test

Statutes or apex-court rulings in major jurisdictions establish machine determination as first instance with human review allocated by screening, quota, or fee, and regulators publish system-level error and disparate-impact audits in place of individual reason-giving.

Disconfirming sign

Apex courts entrench an unrationed individual right to human decision, and administrative systems absorb the cost through caseload backlog rather than delegation.

Themes

Law & institutions, AI & compute, Domestic politics