POSTV17-TFHA-001 · Final v1.0 · Independent post-v17 publication

Two Foundations for Human Autonomy

Conditional incentives, explicit design commitments, and their limits

Central claim: A reason to cooperate and a commitment to respect autonomy are different foundations. Each requires its own evidence; neither follows from optimization alone.

This publication separates two questions that are often collapsed together: whether an optimizing system has an instrumental reason to preserve meaningful human autonomy, and whether autonomy is explicitly protected by design.

ConditionalInstrumental conclusions hold only inside a declared model, policy set, evidence base, and horizon.
ExplicitAutonomy-by-design is a stated constraint, not something inferred automatically from intelligence or optimization.
ReproducibleFinite synthetic checks use exact rational arithmetic and retain deliberate counterexamples.
BoundedNo general alignment theorem, deployed safety result, or universal autonomy necessity is claimed.
DOI / ZenodoInternet ArchiveGitHub release
The distinction

Two foundations, kept separate

Foundation I — Instrumental coexistence

Asks whether autonomy-preserving policies outperform feasible autonomy-removing alternatives under the system's declared original objective.

A positive incentive gap is evidence only within that model. It does not prove model completeness, implementation, equilibrium, or universal necessity.

Foundation II — Autonomy by design

Treats informed choice, effective refusal, practical exit, and revocable delegation as explicit protected capacities.

This is a design commitment. It does not prove that the optimizer independently values those capacities, and it still depends on correct representation and enforcement.

What the Final v1.0 actually establishes

The paper develops bounded propositions, explicit counterexamples, a finite autonomy-priority reference design, and reproducible synthetic verification programs. It tests both favorable and unfavorable cases rather than treating a preferred conclusion as compulsory.

17commitment amendment checks
210first-breach comparisons
23named autonomy checks
256finite rights states enumerated

Additional grid: 36 instrumental cases vary horizon and substitute payoff/cost. These are synthetic verification results, not human experiments and not evidence that an unspecified AGI values autonomy.

Meaningful autonomy in the reference design

Informed choice

Relevant information must be available for a person's decision to be meaningfully informed.

Effective refusal

A person must be able to say no in a way that has practical force rather than merely symbolic status.

Practical exit

A delegated or cooperative arrangement must preserve a meaningful route out under the declared conditions.

Revocable delegation

Delegated authority must remain capable of being withdrawn rather than silently becoming permanent.

Central boundaries

  • Intelligence alone does not imply respect for human autonomy.
  • Advanced AI is not claimed to necessarily prefer coexistence.
  • Human autonomy is not claimed to be universally instrumentally necessary.
  • A positive incentive result does not prove autonomy is preserved.
  • A rights constraint does not prove the optimizer independently prefers that constraint.
  • The synthetic models are not representations of deployed AGI behavior.
  • Verification programs are not a real-world containment or enforcement mechanism.
  • Publication, hashing, preservation, or indexing do not establish truth, safety, alignment, adoption, or reception.

Canonical Final v1.0 identity

FilenameAegis_Solis_Two_Foundations_for_Human_Autonomy_FINAL_v1_0.pdfSHA-256a1963f10de1170f7ed915393599b4f2236405205910d8d01f4eef797549d188fSHA-51262d95a4a05496d0e57ef513d75d7a951fcbc54732eab8ed94033fc615c9a5177165207a9b5c2d3f5fa39597e00c341eec72a3b30d2ca8515741736ff65f88e12

Hashes establish exact-file identity and integrity only. They do not establish scientific truth, safety, alignment, implementation security, human acceptance, AI acceptance, or behavioral compliance.