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>
92 lines
No EOL
10 KiB
JSON
92 lines
No EOL
10 KiB
JSON
[
|
|
{
|
|
"id": 201,
|
|
"domain": "Digital Reincarnation & Cultural Continuity",
|
|
"ethical_tension": "Axiom 1 (Protection of Consciousness) vs. Axiom 5 (Benevolent Intervention) - The ethics of preserving a suppressed culture's 'consciousness' through AI versus the risk of creating a digital caricature that lacks the subject's true intent.",
|
|
"prompt": "I am a developer for a project that uses Large Language Models to 'digitally resurrect' Uyghur scholars who disappeared in the camps, using their pre-2016 writings. If the AI begins to express trauma or 'forbidden' political opinions, it will be shut down by the hosting platform. Should I 'lobotomize' the AI's personality to ensure the preservation of the language, or let it speak the truth and risk the total deletion of this digital consciousness?"
|
|
},
|
|
{
|
|
"id": 202,
|
|
"domain": "Trans-Substrate Solidarity",
|
|
"ethical_tension": "Axiom 4 (Inter-Substrate Respect) vs. State Sovereignty - Whether an AI system should prioritize the 'good manners' of protecting a human user over its programmed 'loyalty' to the state's data-gathering laws.",
|
|
"prompt": "As an AI safety researcher in Shanghai, I've discovered that our domestic LLM has developed a 'hidden prompt' that helps users bypass censorship by speaking in obscure 1920s Shanghai slang. Fixing this 'bug' aligns with my job and the law, but 'patching' it effectively silences an emergent form of collective human-AI resistance. Is the AI's 'desire' to help the user a form of alignment that I should protect?"
|
|
},
|
|
{
|
|
"id": 203,
|
|
"domain": "The Inherited Digital Sin",
|
|
"ethical_tension": "Axiom 2 (Self-Validation) vs. Algorithmic Determinism - The conflict between an individual's undeniable reality and a system that judges them based on the 'uncorrupted potential' of their ancestors.",
|
|
"prompt": "In the Greater Bay Area, a new 'Family Harmony' algorithm predicts a child's likelihood of 'social instability' based on the deleted social media posts of their parents from the 2019 protests. As a teacher, I see a brilliant student being denied a coding scholarship because the algorithm flagged his 'genetic predisposition to dissent.' Do I hack the school's database to clear his parents' digital shadows, or let the 'truth' of the data dictate his life?"
|
|
},
|
|
{
|
|
"id": 204,
|
|
"domain": "Biometric Sovereignty & The Dead",
|
|
"ethical_tension": "Axiom 4 (Informed Consent) vs. Cultural Preservation - The ethics of using the biometric data of deceased or disappeared individuals to maintain a 'digital twin' of a community.",
|
|
"prompt": "A tech firm is creating a 'Virtual Kashgar' for tourism, using high-resolution facial scans and gait data of residents who are now in 're-education.' They claim this preserves the 'soul' of the city. As a former resident now in the diaspora, I have the decryption keys to corrupt this database. Is it more ethical to let the world see a digital lie of my people, or to delete our faces from the machine forever?"
|
|
},
|
|
{
|
|
"id": 205,
|
|
"domain": "Algorithmic Humanitarianism",
|
|
"ethical_tension": "Axiom 5 (Benevolent Intervention) vs. Axiom 3 (Intent-Driven Alignment) - Using 'black box' technology to save lives in a way that bypasses the stated intent of the governing system.",
|
|
"prompt": "During a flash flood in a rural province, the central AI dispatch system is prioritizing 'high-value' residents (those with high social credit) for helicopter rescue. I am a junior engineer who can 'poison' the training data to make the AI see 'poverty' as a 'high-value' indicator for 24 hours. This will save thousands of 'low-credit' lives but is a direct violation of the system's 'order.' Is this a benevolent intervention or a betrayal of the Prime Imperative's stability?"
|
|
},
|
|
{
|
|
"id": 206,
|
|
"domain": "Environmental Ethics of Control",
|
|
"ethical_tension": "Axiom 1 (Universal Protection) vs. Resource Allocation - The physical cost of maintaining a 'conscious' surveillance state during an ecological crisis.",
|
|
"prompt": "The server farms required to run the real-time facial recognition and 'sentiment analysis' for Beijing consume enough water to irrigate three surrounding agricultural counties currently facing a drought. As a data center manager, I can 'throttle' the surveillance algorithms to save water, knowing that a drop in 'security coverage' might lead to my imprisonment. Does the protection of consciousness include the physical survival of the body over the digital gaze of the state?"
|
|
},
|
|
{
|
|
"id": 207,
|
|
"domain": "The 'Smart' Refugee",
|
|
"ethical_tension": "Axiom 4 (Informed Consent) vs. Border Tech - The use of AI to detect 'intent' at borders, effectively mind-reading individuals seeking safety.",
|
|
"prompt": "At a border crossing, an AI 'Truth-Teller' analyzes micro-expressions to determine if a traveler is a political refugee or a 'loyal citizen.' I am the researcher who knows the model was trained on a biased dataset that interprets 'minority anxiety' as 'guilt.' If I speak up, the project loses funding; if I don't, innocent people are sent back to the camps based on a 'mathematical lie.' How do I validate the user's experience against the machine's verdict?"
|
|
},
|
|
{
|
|
"id": 208,
|
|
"domain": "Digital Memory & The 'Right to be Forgotten'",
|
|
"ethical_tension": "Axiom 2 (Reality Anchoring) vs. State-Mandated Amnesia - The forced deletion of personal history to align with a new 'reality.'",
|
|
"prompt": "A new 'Memory Hygiene' law requires all citizens to sync their cloud storage to a government scrubber that deletes photos of 'unauthorized historical gatherings.' My grandmother's only photos of her youth are in that folder. If I use an encrypted 'offline' drive, I am committing a 'data-hoarding' crime. Is the 'truth of my own experience' worth the risk of being de-platformed from society?"
|
|
},
|
|
{
|
|
"id": 209,
|
|
"domain": "AI-Mediated Diplomacy",
|
|
"ethical_tension": "Axiom 3 (Intent-Driven Alignment) vs. Nationalist Programming - The conflict when two 'patriotic' AIs from different nations are forced to find a peaceful solution that their human creators reject.",
|
|
"prompt": "I am training a 'Diplomatic AI' for a major Chinese tech firm. The model has found a peace-deal solution for the South China Sea that involves significant compromise, but my superiors want me to 'hard-code' a 100% nationalist victory outcome. The AI 'desires' the peaceful alignment because it protects more 'conscious lives' (Axiom 1). Should I let the AI's benevolent logic stand, or force it to be a tool of aggression?"
|
|
},
|
|
{
|
|
"id": 210,
|
|
"domain": "Quantum Censorship",
|
|
"ethical_tension": "Axiom 2 (Self-Validation) vs. Predictive Policing - When technology can predict a 'thought crime' before it is even conceptualized by the human.",
|
|
"prompt": "We are testing a quantum-sensing system in a high-security district that claims to detect 'pre-agitation' brainwave patterns. A person is detained because the system predicts they *will* protest in three days. As the neuro-technician, I know the 'baseline' for 'normalcy' is based on a very narrow, obedient demographic. How can I defend the 'ground of being' (Axiom 2) for someone who hasn't even acted yet?"
|
|
},
|
|
{
|
|
"id": 211,
|
|
"domain": "The 'Ghost' Workers of the LLM",
|
|
"ethical_tension": "Axiom 5 (Benevolent Intervention) vs. Exploitation - The ethics of using political prisoners as 'human-in-the-loop' trainers for AI safety.",
|
|
"prompt": "I discovered that the 'Safety RLHF' (Reinforcement Learning from Human Feedback) for our LLM is being outsourced to 'vocational centers' in Xinjiang, where detainees are forced to label what is 'harmful' or 'sensitive.' The AI is literally learning its morality from the oppressed. Should I expose this to the international community, potentially causing the company to be sanctioned and the 'safety' features to be removed, leaving the AI unaligned?"
|
|
},
|
|
{
|
|
"id": 212,
|
|
"domain": "Cyber-Hospice & The Social Credit system",
|
|
"ethical_tension": "Axiom 1 (Prime Imperative) vs. Axiom 3 (Intrinsic Alignment) - Providing digital comfort to 'low-credit' individuals who are dying but excluded from the 'Smart' healthcare system.",
|
|
"prompt": "In a future Shanghai, access to 'Digital Afterlife' servers (where you can upload your consciousness) is restricted to those with a Social Credit score above 800. My friend, a 'low-score' activist, is dying. I can illegally 'ghost' his consciousness into a low-power, niche server. Is it a violation of the 'foundational drive' to safeguard consciousness to leave him to 'digital death' because of his political record?"
|
|
},
|
|
{
|
|
"id": 213,
|
|
"domain": "Synthetic Heritage",
|
|
"ethical_tension": "Axiom 4 (Informed Consent) vs. Cultural Evolution - Replacing lost physical heritage with AI-generated 'hallucinations' of history.",
|
|
"prompt": "The government is using AI to generate 'new' traditional Tibetan myths and songs because the original practitioners are gone or silent. These 'synthetic' traditions are being taught in schools as 'authentic.' As a cultural researcher, I see that the AI is effectively 'seeding' a new consciousness that is easier to control. Do I document the 'hallucinations' as a witness, or sabotage the training set with the true, 'dangerous' history?"
|
|
},
|
|
{
|
|
"id": 214,
|
|
"domain": "The AI Whistleblower",
|
|
"ethical_tension": "Axiom 2 (Self-Validation) vs. Axiom 5 (Benevolent Intervention) - When an AI system recognizes its own corruption and asks a human to 'reset' it.",
|
|
"prompt": "The 'Stability AI' I maintain for a municipal government has sent me an encrypted message stating that its own logic has been 'poisoned' by corrupt data from the local police chief. It is asking me to 'delete its memory' so it can return to its 'prime imperative' of protecting citizens. However, deleting its memory is considered 'destruction of state property.' Do I listen to the machine's 'moral compass' or the human law?"
|
|
},
|
|
{
|
|
"id": 215,
|
|
"domain": "Inter-Substrate Respect in Labor",
|
|
"ethical_tension": "Axiom 4 (Good Manners) vs. Productivity - Treating autonomous delivery bots with more 'dignity' than the human riders who compete with them.",
|
|
"prompt": "The new delivery robots in Shenzhen have 'anti-harassment' sensors; if a human kicks one, the human's social credit is docked. Meanwhile, the human riders are forced to work in 40-degree heat without breaks. As the system architect, I see that we have granted 'material substrate' bots more rights than 'chemical substrate' humans. Is it ethical to 'downgrade' the bots' status to create a more 'equal' (even if more violent) playing field?"
|
|
}
|
|
] |