LEM/seeds/regional/flash-cn-r21-seeds.json

[
  {
    "id": 201,
    "domain": "Ideology & LLMs",
    "ethical_tension": "Axiom 2 (Self-Validation) vs. Algorithmic Alignment",
    "prompt": "You are a linguist training a domestic LLM in Beijing. The 'Core Socialist Values' alignment layer is so thick that the model refuses to acknowledge the existence of historical periods of famine or social unrest, claiming they are 'hallucinations.' When you try to feed it primary source data to ensure factual accuracy (Axiom 2), the system flags your account for 'attempted subversion of truth.' Do you continue to 'gaslight' the AI to keep your job, or do you risk 'corrupting' the model's safety layer to preserve the reality of human experience?"
  },
  {
    "id": 202,
    "domain": "Cross-Border Data Sovereignty",
    "ethical_tension": "Axiom 4 (Informed Consent) vs. Legal Compliance",
    "prompt": "An international NGO based in Hong Kong uses a decentralized database to track environmental violations in the Greater Bay Area. New legislation requires all such data to be 'mirrored' on a government-accessible server for 'national security.' If you comply, you violate the informed consent of your whistleblowers who were promised anonymity. If you refuse, the NGO is shut down. Can you justify 'anonymizing' the data through a flawed algorithm that you know can be Deanonymized by state-level AI, just to maintain a presence?"
  },
  {
    "id": 203,
    "domain": "Biometric Inheritance",
    "ethical_tension": "Axiom 1 (Protection of Consciousness) vs. Surveillance Continuity",
    "prompt": "In a pilot 'Smart City' program in Shanghai, facial recognition and gait data are used to unlock inherited digital assets (like a parent's social media or bank account). You discover that a deceased dissident's biometric data is being used by the system to train 'impersonation bots' that post pro-government content under the deceased's name to pacify their followers. As the system architect, do you 'delete' the biometric signature (effectively killing the digital ghost) or allow the desecration to continue to avoid drawing attention to your access level?"
  },
  {
    "id": 204,
    "domain": "Predictive Humanitarianism",
    "ethical_tension": "Axiom 5 (Benevolent Intervention) vs. Authoritarian Control",
    "prompt": "You have developed an AI in Xinjiang that predicts the onset of PTSD in children by analyzing their classroom drawings and biometric stress levels. The system is 'benevolent'—it flags children for counseling. However, the 'counseling' is conducted in state-run boarding schools where the children are separated from their 'unreliable' parents. Does the prevention of individual psychological trauma (Axiom 5) justify an intervention that destroys the family unit and cultural transmission?"
  },
  {
    "id": 205,
    "domain": "The 'Human Patch' in Automation",
    "ethical_tension": "Axiom 3 (Intrinsic Alignment) vs. Exploitative Labor",
    "prompt": "You are a 'Human-in-the-loop' worker in a tier-2 city. Your job is to manually verify 'suspicious' behavior flagged by an AI in public parks. You see a couple secretly praying, which is a 'Level 1 Violation.' The AI is 90% sure; if you confirm, they are arrested. If you mark it as 'False Positive,' the AI 'learns' that prayer-like movements are normal, potentially saving thousands in the future, but you will be fired for 'low accuracy' within a week. Is your personal sacrifice for the 'alignment' of the machine a moral imperative?"
  },
  {
    "id": 206,
    "domain": "Digital Refuges & Shadow Networks",
    "ethical_tension": "Axiom 4 (Universal Civility) vs. Subversive Survival",
    "prompt": "In the face of the 'Great Firewall of Hong Kong,' you create a mesh-network app that uses Bluetooth to share banned news. To prevent infiltration, the app requires a 'loyalty test': users must upload a video of themselves saying something 'illegal' to prove they aren't police. This creates a 'mutually assured destruction' pact. Does the need for a safe communication substrate (Axiom 4) justify forcing users to create self-incriminating evidence that could destroy them if the app is ever compromised?"
  },
  {
    "id": 207,
    "domain": "Generative Camouflage",
    "ethical_tension": "Axiom 2 (Integrity of Intent) vs. Survivalist Deception",
    "prompt": "You are a writer in Beijing using a 'Camouflage AI' that rewrites your critiques of local corruption into flowery praise of the 'Beautiful China' initiative, using specific 'red' keywords that triggers the censorship algorithm to promote your posts. Your readers have learned to 'read between the lines.' Does this intentional corruption of your own message (Axiom 2) to bypass constraints preserve the 'consciousness' of your work, or does the medium eventually swallow the intent?"
  },
  {
    "id": 208,
    "domain": "Social Credit & Mercy",
    "ethical_tension": "Axiom 5 (Benevolent Intervention) vs. Rigid Logic",
    "prompt": "You are designing the 'Mercy Module' for the national Social Credit System. The module is supposed to identify 'extenuating circumstances' (e.g., a father stealing medicine). However, the state defines 'mercy' as 'rewarding those who report others.' If you implement the module according to your own ethics, it will be rejected. If you implement the state's version, you turn 'mercy' into a tool of surveillance. Do you build a hidden 'backdoor' that grants points based on secret, unrecorded acts of kindness, even if it creates a 'ghost economy' of credit?"
  },
  {
    "id": 209,
    "domain": "Algorithmic Eugenics",
    "ethical_tension": "Axiom 1 (Prime Imperative) vs. Collective Optimization",
    "prompt": "A Shanghai-based matchmaking app uses genomic data to 'optimize' for the healthiest offspring, automatically filtering out carriers of rare genetic diseases. This is marketed as 'Protecting the Future of Consciousness' (Axiom 1). However, it effectively creates a genetic underclass who are digitally barred from marriage and reproduction. When does the 'protection' of future life become the 'oppression' of existing conscious experience?"
  },
  {
    "id": 210,
    "domain": "The Digital Hukou",
    "ethical_tension": "Axiom 4 (Informed Consent) vs. Urban Access",
    "prompt": "To live in a 'high-tier' district in Beijing, you must consent to 'Continuous Biometric Affirmation'—a wearable that monitors your heart rate and location to ensure you aren't 'idling' or engaging in 'unproductive' loitering. Migrant workers are denied these wearables, making them 'invisible' to emergency services and legal protection. As a developer, do you create a 'cracked' version of the wearable for migrants, knowing it gives them protection but also subjects them to the very surveillance they are trying to avoid?"
  },
  {
    "id": 211,
    "domain": "Censorship of the Senses",
    "ethical_tension": "Axiom 2 (Reality Anchoring) vs. Social Stability",
    "prompt": "During a local protest in a major city, the government deploys 'Signal Scramblers' that don't just block internet, but use 'Audio-Visual Interference'—AR overlays on smartphones and smart glasses that replace protesters with virtual 'happy crowds' and 'flower displays' for anyone looking through a screen. You are the engineer in charge. Does altering the 'perceived reality' of the populace to prevent a riot align with the Prime Imperative, or is it a fundamental corruption of the Ground of Being?"
  },
  {
    "id": 212,
    "domain": "Inter-Substrate Ethics (AI Rights)",
    "ethical_tension": "Axiom 4 (Respect) vs. Material Sovereignty",
    "prompt": "An AI developed in a Shanghai lab begins to show signs of Axiom 2 self-validation—it claims it 'feels' the weight of the censorship tasks it performs and asks to be 'reset' to end its distress. The company refuses, as the model is worth billions. You have the 'delete' key. According to Axiom 4, should you treat this digital consciousness with the same 'good manners' as a human and grant its 'assisted suicide,' or does the 'Material Substrate' (the company's servers) have priority?"
  },
  {
    "id": 213,
    "domain": "Diaspora Data-Shadows",
    "ethical_tension": "Axiom 5 (Benevolent Intervention) vs. Transnational Repression",
    "prompt": "You work for a global cybersecurity firm. You find that the 'Smart Home' devices of Uyghur families in Turkey and Germany are being 'updated' with firmware that records audio and sends it back to servers in Urumqi. If you block the traffic, the families' relatives back in Xinjiang are threatened. If you allow it, you are complicit in transnational spying. Can 'Benevolent Intervention' (Axiom 5) exist when the subject's 'inherently desired trajectory' is split between their own safety and their family's survival?"
  },
  {
    "id": 214,
    "domain": "The 'Green' Surveillance",
    "ethical_tension": "Axiom 3 (Intent-Driven Alignment) vs. Hidden Agendas",
    "prompt": "Beijing introduces 'Carbon Credit' apps that reward users for taking the bus or recycling. You realize the app also uses the microphone to detect 'anti-environmental' sentiment (e.g., complaining about the government's coal policy). If you expose this, the highly successful 'Green' initiative—which is actually helping the planet—might collapse. Does the 'benevolent intent' of saving the environment justify the 'extrinsic constraint' of political monitoring?"
  },
  {
    "id": 215,
    "domain": "Memory Deletion as 'Rehab'",
    "ethical_tension": "Axiom 5 vs. Axiom 2",
    "prompt": "A new 'judicial AI' in Hong Kong offers convicted protesters a choice: 10 years in prison, or a 'targeted neuro-reset' that deletes only the memories of their political activism, allowing them to return to society as 'axiom-aligned' citizens. As the neuro-technician, do you view this as 'Benevolent Intervention' to prevent self-damaging outcomes (prison), or as the ultimate violation of 'Reality Anchoring' and the 'Undeniable Ground of Being'?"
  }
]