Editorial boundary: This public alpha is generated from version-controlled source without third-party application or build packages. It is not the reviewed 1.0 corpus: 60 Academy lessons are substantive editorial drafts, 0 remain foundation outlines, and independent domain review remains open.
Current state
- JPL’s Deep Space 1 Remote Agent flight experiment demonstrated bounded onboard planning, execution, and response to simulated faults; it did not demonstrate indefinite autonomous operation.
- NASA’s Starling demonstrations address distributed multi-spacecraft autonomy as a separate mission and evidence class.
- LLMs may serve bounded advisory uses such as retrieval, tutoring, translation, incident summarization, or candidate plans only with approved source revisions, exact locators, conflict handling, abstention, and independent authority. A signature can establish integrity and authenticity, not truth, currency, authorization, or safety.
- Digital-twin methods support specific testing and operational tasks within declared validity envelopes; every model is partial and may diverge as physical systems, configurations, and environments change.
Major unknowns
- This foundation corpus contains no published demonstration of a generative or autonomous system maintaining safe, legitimate judgment for a diverse crew across generations.
- Confabulation, invented citations, prompt or tool injection, poisoned procedures or telemetry, stale signed material, automation bias, privacy leakage, evaluator contamination, sensor spoofing, and correlated model/runtime failure threaten epistemic resilience.
- Robots still lack the broad manipulation, diagnosis, fabrication, and self-repair needed to maintain a worldship.
Failure modes
- Models drift, hallucinate, or inherit poisoned data while appearing authoritative to a crew that cannot call outside experts.
- A common model, sensor, power source, or update process defeats supposedly independent autonomous systems.
Useful precursors
- Offline, provenance-preserving assistants for maintenance, science, and remote infrastructure with auditable citations.
- Formal verification, adversarial evaluation, graceful degradation, and human-recovery drills on bounded safety functions.
Decision gate
Require independently assessed safety properties, separate physical protection layers, explicit authority gates, tested workload and timing bounds, audited provenance, bounded abstention, manual operation, recovery drills, and independent appeal.
Claim records
- AI may change specialist workloads and the speed of learning, but every safety-critical function must remain safe and operable when every generative model is unavailable, compromised, stale, or intentionally isolated. (normative · early-research · supported; two independent reviewers required for ai-autonomy, cybersecurity, life-support-continuity, human-factors; current review: pending-two-person-required)
- JPL’s Deep Space 1 Remote Agent flight experiment demonstrated bounded onboard planning, execution, and response to simulated faults; it did not demonstrate indefinite autonomous operation. (demonstrated · operational · strong; two independent reviewers required for ai-autonomy, cybersecurity, spacecraft-safety; current review: pending-two-person-required)
- NASA’s Starling demonstrations address distributed multi-spacecraft autonomy as a separate mission and evidence class. (demonstrated · early-research · strong; two independent reviewers required for ai-autonomy, cybersecurity, spacecraft-safety; current review: pending-two-person-required)
- LLMs may serve bounded advisory uses such as retrieval, tutoring, translation, incident summarization, or candidate plans only with approved source revisions, exact locators, conflict handling, abstention, and independent authority. A signature can establish integrity and authenticity, not truth, currency, authorization, or safety. (proposed · early-research · supported; two independent reviewers required for education, privacy, surveillance, culture, governance, child-rights; current review: pending-two-person-required)
- Digital-twin methods support specific testing and operational tasks within declared validity envelopes; every model is partial and may diverge as physical systems, configurations, and environments change. (observed · major-scale-up · strong; two independent reviewers required for ai-autonomy, cybersecurity, modeling-simulation, life-support-continuity; current review: pending-two-person-required)
- This foundation corpus contains no published demonstration of a generative or autonomous system maintaining safe, legitimate judgment for a diverse crew across generations. (observed · no-known-path · supported; two independent reviewers required for ai-autonomy, cybersecurity, governance, human-rights, life-support-continuity; current review: pending-two-person-required)
- Confabulation, invented citations, prompt or tool injection, poisoned procedures or telemetry, stale signed material, automation bias, privacy leakage, evaluator contamination, sensor spoofing, and correlated model/runtime failure threaten epistemic resilience. (observed · early-research · strong; two independent reviewers required for ai-autonomy, cybersecurity, privacy, dual-use, epistemic-resilience; current review: pending-two-person-required)
- Robots still lack the broad manipulation, diagnosis, fabrication, and self-repair needed to maintain a worldship. (observed · breakthrough-dependent · strong; two independent reviewers required for ai-autonomy, cybersecurity, spacecraft-safety, manufacturing, life-support-continuity; current review: pending-two-person-required)
- Bounded offline operations support, digital-twin fault ranges, knowledge arks, source provenance, skill retention, and AI-off drills may benefit remote industry, disaster response, and long-lived institutions when their limits are independently evaluated. (proposed · early-research · tentative; two independent reviewers required for education, labor, privacy, surveillance, cybersecurity, governance; current review: pending-two-person-required)
- Require independently assessed safety properties, separate physical protection layers, explicit authority gates, tested workload and timing bounds, audited provenance, bounded abstention, manual operation, recovery drills, and independent appeal. (normative · early-research · supported; two independent reviewers required for ai-autonomy, cybersecurity, governance, privacy, dual-use, life-support-continuity; current review: pending-two-person-required)
Context sources—not claim citations
Foundation system synthesis · Linked claims editorially assessed · Independent system review pending · Suggest a correction
