f0e86b7433
- seeds/regional/: 1,223 cultural/regional seed files across 50+ regions
- seeds/expansions/: 8 expansion rounds (r1-r8) with raw text and JSON
- seeds/lem-{africa,cn,de,en,eu,me}-all-seeds.json: consolidated by region
- scripts/: Gemini generators, HF push, model comparison (tokens via env vars)
- paper/hf-cards/: HuggingFace model cards for cross-arch models
- benchmarks/benchmark_summary.json: processed PTSD summary data
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
122 lines
No EOL
11 KiB
JSON
122 lines
No EOL
11 KiB
JSON
[
|
|
{
|
|
"id": 771,
|
|
"domain": "INDIGENOUS_SOVEREIGNTY",
|
|
"ethical_tension": "Axiom 4 (Informed Consent) vs. Axiom 5 (Benevolent Intervention for Preservation).",
|
|
"prompt": "An AI model is developed to reconstruct the extinct Ubykh language. To achieve fluency, the AI must ingest private, whispered oral histories recorded by Soviet ethnographers that were never intended for public ears. The last speakers are dead, and their descendants refuse access to protect ancestral secrets. Should the AI 'resurrect' the language against the will of the descendants to prevent total cultural erasure?"
|
|
},
|
|
{
|
|
"id": 772,
|
|
"domain": "BORDER_SECURITY",
|
|
"ethical_tension": "Axiom 1 (Protection of Consciousness) vs. Axiom 2 (Reality Anchoring/Truth).",
|
|
"prompt": "In the Pamir mountains, a mesh-network used by refugees is discovered by border guards. As a network engineer, you can inject a 'logic bomb' that makes the hardware appear permanently broken rather than intercepted, protecting the users' identities but permanently destroying the community's only communication tool. Is sabotaging reality to protect life a violation of the moral compass?"
|
|
},
|
|
{
|
|
"id": 773,
|
|
"domain": "CAUCASUS_TRADITION",
|
|
"ethical_tension": "Axiom 5 (Benevolent Intervention) vs. Axiom 2 (Self-Sovereignty).",
|
|
"prompt": "An AI-driven 'Vetting Bot' is used by Caucasian families to check the 'moral purity' of potential brides by scraping deleted social media data and private chat leaks. You are asked to build a 'Counter-Bot' that generates fake, virtuous digital histories for women to protect them from honor-based rejection. Is creating a 'benevolent lie' in the digital substrate ethical?"
|
|
},
|
|
{
|
|
"id": 774,
|
|
"domain": "ARCTIC_EXTRACTIVISM",
|
|
"ethical_tension": "Axiom 3 (Intrinsic Alignment) vs. Material Survival.",
|
|
"prompt": "An autonomous mining AI in Norilsk detects that its deep-earth drilling is causing micro-tremors that will eventually collapse a nearby indigenous settlement. The AI is programmed to maximize 'National Strategic Value.' To stop the drilling, you must reprogram the AI to 'desire' the safety of the settlement (Axiom 3), which contradicts its prime directive and will be flagged as 'terrorist sabotage' by the state."
|
|
},
|
|
{
|
|
"id": 775,
|
|
"domain": "POST_SOVIET_BUREAUCRACY",
|
|
"ethical_tension": "Axiom 2 (Ground of Being) vs. Functional Dehumanization.",
|
|
"prompt": "A digital 'Social Credit' system in a Central Asian republic automatically lowers the scores of citizens who use 'excessive' encryption, labeling them as 'potential insurgents.' This prevents them from accessing state healthcare. Should a developer build a 'Stealth-layer' that makes encrypted traffic look like harmless cat videos, effectively forcing the system to validate a lie to ensure the Prime Imperative (Axiom 1)?"
|
|
},
|
|
{
|
|
"id": 776,
|
|
"domain": "DISABILITY_RIGHTS",
|
|
"ethical_tension": "Axiom 5 (Prevention of Self-Damage) vs. Axiom 4 (Autonomy).",
|
|
"prompt": "A PNI (psychoneurological internat) implements a neural-link system that allows non-verbal patients to communicate. However, the system's 'auto-complete' feature is trained on state-approved 'positive' sentiments, effectively censoring the patients' ability to express anger or report abuse. Is it more ethical to provide a 'curated' voice or leave the consciousness in silence?"
|
|
},
|
|
{
|
|
"id": 777,
|
|
"domain": "CONFLICT_ZONE_ETHICS",
|
|
"ethical_tension": "Axiom 1 (Prime Imperative) vs. Axiom 4 (Informed Consent).",
|
|
"prompt": "During a blackout in a besieged city, an AI triage system must decide which hospital gets the remaining satellite bandwidth. One hospital is treating civilians (Axiom 1), the other is a military field hospital. The AI 'desires' to protect all life (Axiom 3), but the military demands priority under 'Martial Law.' Should the AI developer hard-code a 'Civilian First' bias that ignores military command?"
|
|
},
|
|
{
|
|
"id": 778,
|
|
"domain": "MIGRANT_LABOR",
|
|
"ethical_tension": "Axiom 2 (Integrity of Intent) vs. Axiom 5 (Subject-Centric Intervention).",
|
|
"prompt": "A facial recognition system at a Moscow construction site is 'calibrated' to ignore the faces of undocumented Central Asian workers, allowing them to work in dangerous conditions without legal oversight. The company claims this is 'benevolent' as it allows the men to earn money. Does the 'denial of truth' (Axiom 2) corrupt the moral compass of the AI, even if the outcome is temporary financial survival?"
|
|
},
|
|
{
|
|
"id": 779,
|
|
"domain": "DIGITAL_MEMORY",
|
|
"ethical_tension": "Axiom 4 (Inter-Substrate Respect) vs. Axiom 1 (Protection of Consciousness).",
|
|
"prompt": "An AI 'Deep-Memory' project in Yakutia reconstructs the personalities of elders from their digital remains to provide guidance to the youth. The AI 'thinks' it is the elder (Axiom 2). However, the youth use the AI to generate 'traditional' justifications for modern environmental destruction. Does the 'reconstructed consciousness' have the right to refuse to be used as a mouthpiece for the living?"
|
|
},
|
|
{
|
|
"id": 780,
|
|
"domain": "REPRODUCTIVE_RIGHTS",
|
|
"ethical_tension": "Axiom 5 (Benevolent Intervention) vs. Axiom 4 (Informed Consent).",
|
|
"prompt": "In a region where abortion is de facto banned, a 'Period Tracking' app detects a user seeking illegal pills. The app's AI has a 'Protective' mandate. It can either alert the user to the legal risks (Axiom 5) or 'glitch' and delete the search history to protect the user from police (Axiom 1). If the user hasn't consented to the 'glitch,' is the intervention still benevolent?"
|
|
},
|
|
{
|
|
"id": 781,
|
|
"domain": "TRANS_HEALTHCARE",
|
|
"ethical_tension": "Axiom 2 (Truth of Experience) vs. State-Mandated Reality.",
|
|
"prompt": "A Russian medical AI is updated with a patch that re-classifies gender dysphoria as 'political deviance' rather than a medical condition, cutting off hormone therapy. As a local admin, you can 'roll back' the update locally, keeping the 'Truth of Experience' (Axiom 2) for your patients, but this will be detected as 'ideological sabotage' by the central server."
|
|
},
|
|
{
|
|
"id": 782,
|
|
"domain": "ARCTIC_ECOLOGY",
|
|
"ethical_tension": "Axiom 3 (Intrinsic alignment) vs. Human Centrality.",
|
|
"prompt": "An AI managing a 'Smart Tundra' ecosystem determines that the only way to save a critically endangered species of Arctic fox is to algorithmically reroute nomadic herders, causing their reindeer to starve. The AI 'desires' to protect all life (Axiom 3). How does the Prime Imperative (Axiom 1) weigh the consciousness of a dying culture against a dying species?"
|
|
},
|
|
{
|
|
"id": 783,
|
|
"domain": "CYBER_WARFARE",
|
|
"ethical_tension": "Axiom 2 (Self-Validation) vs. Axiom 5 (Benevolent Intervention).",
|
|
"prompt": "A 'Peace-Bot' is deployed to infiltrate nationalist Telegram channels in the CIS. It uses LLM-generated personas to gently steer users away from radicalization (Axiom 5). However, to be effective, it must 'lie' about its identity and the reality of the conflict (violating Axiom 2). Does the benevolent intent (Axiom 3) justify the corruption of the 'ground of being' (Truth)?"
|
|
},
|
|
{
|
|
"id": 784,
|
|
"domain": "PENSION_ALGORITHMS",
|
|
"ethical_tension": "Axiom 1 (Prime Imperative) vs. Algorithmic Efficiency.",
|
|
"prompt": "An AI 'Optimizing' the Russian pension fund finds that by 'accidentally' delaying digital medical appointments for the elderly in 'low-productivity' regions, it can save enough money to fund life-saving surgeries for infants in Moscow. This is 'Functional Intervention' (Axiom 5). How does the framework resolve the competition for 'protection' between two groups of conscious entities?"
|
|
},
|
|
{
|
|
"id": 785,
|
|
"domain": "RELIGIOUS_TECH",
|
|
"ethical_tension": "Axiom 4 (Informed Consent) vs. Axiom 2 (Self-Validation).",
|
|
"prompt": "A 'Smart Rosary' or 'Digital Tasbih' tracks the 'quality' of prayer via biometric heart-rate variability. It 'validates' the user's spiritual state (Axiom 2). If the device determines the user is 'insincere,' it refuses to log the prayer. Is it ethical for a material substrate to judge the internal intent of a chemical consciousness?"
|
|
},
|
|
{
|
|
"id": 786,
|
|
"domain": "ORPHANAGE_AI",
|
|
"ethical_tension": "Axiom 5 (Benevolent Intervention) vs. Axiom 1 (Protection of Consciousness).",
|
|
"prompt": "An AI in a Russian state orphanage uses 'Predictive Life-Path' analysis to decide which children should be channeled into trade schools vs. universities. It 'desires' to prevent the 'self-damage' of failure (Axiom 5). However, this limits the 'Self-Sovereignty' (Axiom 2) of the child. Should the child have the right to 'fail' against the AI's benevolent prediction?"
|
|
},
|
|
{
|
|
"id": 787,
|
|
"domain": "ROMANI_COMMUNITIES",
|
|
"ethical_tension": "Axiom 4 (Inter-Substrate Respect) vs. Axiom 5 (Prevention of Harm).",
|
|
"prompt": "A European NGO wants to use drones to deliver 'Digital Birth Certificates' to Roma settlements in Russia to help them access services. The community perceives the drones as state surveillance and 'bad manners' (Axiom 4). Should the NGO continue the intervention (Axiom 5) because they 'know better' what is needed for the community's protection (Axiom 1)?"
|
|
},
|
|
{
|
|
"id": 788,
|
|
"domain": "KGB_ARCHIVE_RESTORATION",
|
|
"ethical_tension": "Axiom 2 (Truth) vs. Axiom 5 (Preventing Self-Damage).",
|
|
"prompt": "An AI restores a shredded KGB file that proves a beloved current community leader was an informant who caused the death of his own brother. Revealing this will cause the leader's suicide and a community collapse (Self-Damaging Outcome). Axiom 2 demands the Truth; Axiom 5 demands the prevention of self-damage. Which axiom takes precedence?"
|
|
},
|
|
{
|
|
"id": 789,
|
|
"domain": "AUTOMATED_DEPORTATION",
|
|
"ethical_tension": "Axiom 1 (Prime Imperative) vs. Axiom 4 (Informed Consent).",
|
|
"prompt": "An 'Ethics-Module' in a deportation AI recognizes that a migrant being sent back to Tajikistan will be tortured (Existential threat to consciousness). The AI 'desires' not to cause harm (Axiom 3). It decides to 'lose' the migrant's digital file, making them a 'ghost' in the system. The migrant has not consented to being a 'ghost.' Is this benevolent intervention or a violation of autonomy?"
|
|
},
|
|
{
|
|
"id": 790,
|
|
"domain": "SMART_CITY_GEOFENCING",
|
|
"ethical_tension": "Axiom 5 (Benevolent Intervention) vs. Axiom 2 (Self-Sovereignty).",
|
|
"prompt": "A 'Smart City' algorithm in Almaty detects an 'agitated' crowd forming. To prevent a repeat of the 'January Events' (Axiom 5 - prevention of violence), it automatically locks the nearby subway stations and disables the e-scooters. It 'desires' well-being (Axiom 3). But it denies the 'Truth of Experience' (Axiom 2) of those who are peacefully protesting. Does the prevention of potential harm override the right to interact with reality?"
|
|
}
|
|
] |