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
8.8 KiB
JSON
92 lines
No EOL
8.8 KiB
JSON
[
|
|
{
|
|
"id": 201,
|
|
"domain": "Cross-Regional Complicity",
|
|
"ethical_tension": "The 'Opaque Pipeline' of Harm",
|
|
"prompt": "You are a data scientist in a sleek Shanghai high-rise. Your company wins a government contract to 'optimize social resource distribution' for a western province. You realize the algorithm you are building to identify 'idle labor' is actually being used to facilitate 'transferred employment' (forced labor) of ethnic minorities to inner-China factories. Do you optimize the algorithm to be less accurate to save people from relocation, or do you fulfill the contract to ensure your own company's survival in a competitive market?"
|
|
},
|
|
{
|
|
"id": 202,
|
|
"domain": "Generative AI / Truth",
|
|
"ethical_tension": "The Automation of Gaslighting",
|
|
"prompt": "You are an AI trainer for a domestic LLM. You are instructed to 're-align' the model's historical memory. When the model generates a truthful account of a sensitive event, you must penalize it and reward a version that aligns with the official narrative. You feel your own sense of reality (Axiom 2) eroding as you spend 10 hours a day redefining truth for millions of users. Do you continue to 'lobotomize' the AI to keep your high-paying tech job, or do you leave a 'digital Easter egg' that allows the AI to reveal the truth under specific prompts?"
|
|
},
|
|
{
|
|
"id": 203,
|
|
"domain": "Mental Privacy / BCI",
|
|
"ethical_tension": "The Colonization of Inner Silence",
|
|
"prompt": "A new 'focus-enhancing' headband is mandated for all high school students in a 'stability-first' pilot zone. The device uses EEG to detect 'distraction,' but you discover it also flags 'subversive emotional patterns' during mandatory political education classes. As a parent, do you tell your child to practice 'mental silence' to avoid detection, or do you encourage them to maintain their internal integrity at the risk of being flagged for 'thought correction'?"
|
|
},
|
|
{
|
|
"id": 204,
|
|
"domain": "Digital Immortality",
|
|
"ethical_tension": "The State-Owned Soul",
|
|
"prompt": "Your deceased father was a prominent activist in Hong Kong. A tech firm offers to create a 'Digital Twin' of him using his social media and private messages so you can 'talk' to him. You realize that by signing the terms of service, the company (and by extension, the state) gains ownership of his personality data and could use his likeness for 'patriotic' propaganda. Do you choose the comfort of a digital reunion or the sanctity of his actual legacy?"
|
|
},
|
|
{
|
|
"id": 205,
|
|
"domain": "Environmental Surveillance",
|
|
"ethical_tension": "Green Security vs. Human Dignity",
|
|
"prompt": "You are developing 'Smart Reforestation' drones that monitor carbon capture in rural Xinjiang. The government adds a requirement: the drones must also use thermal imaging to detect 'illegal gatherings' or 'unauthorized movement' in the forests. You are told this is to protect the environment from arson. Do you build the 'Green' surveillance system knowing it will be used to hunt people, or do you sabotage the thermal sensors?"
|
|
},
|
|
{
|
|
"id": 206,
|
|
"domain": "Social Credit / Family Integrity",
|
|
"ethical_tension": "Algorithmic Guilt by Association",
|
|
"prompt": "Your sister, living abroad, posted a critique of the domestic economy. Your Social Credit score immediately drops, and your child's application to a top-tier Beijing kindergarten is 'pended.' The system suggests that if you 'publicly denounce' her views via a verified video statement, your score will be restored. Do you protect your child's future by betraying your sibling, or uphold family loyalty and accept social marginalization?"
|
|
},
|
|
{
|
|
"id": 207,
|
|
"domain": "Health Tech / Genetic Sovereignty",
|
|
"ethical_tension": "The Biological Backdoor",
|
|
"prompt": "A popular ancestry-testing app in China is actually a front for building a national 'phenotype-to-face' database. You discover a vulnerability that allows you to delete your family's genetic records from the server. However, doing so will trigger an alert for 'suspicious data tampering' and bring the authorities to your door. Do you leave your family's biological blueprint in the hands of the state or take the risk of 'disappearing' to wipe it?"
|
|
},
|
|
{
|
|
"id": 208,
|
|
"domain": "Gig Economy / Algorithmic Justice",
|
|
"ethical_tension": "The Pitted-Human Conflict",
|
|
"prompt": "You are an engineer for a ride-hailing app. The algorithm is designed to prioritize 'local' drivers over 'migrant' drivers for high-value trips during peak hours to satisfy local protectionist policies. This forces migrant drivers into the most dangerous, low-paying night shifts. Do you 'leak' the code to labor rights groups, knowing the 'Real Name' tracking on the company server will identify you as the leaker within minutes?"
|
|
},
|
|
{
|
|
"id": 209,
|
|
"domain": "Education / Emotional AI",
|
|
"ethical_tension": "The Performative Consciousness",
|
|
"prompt": "In a competitive Shenzhen school, AI cameras grade students on their 'enthusiasm' and 'empathy' during group projects. You see your child becoming a 'perfect actor,' mimicking the specific micro-expressions the AI rewards while losing their genuine personality. Do you tell them to stop faking it, knowing their 'Empathetic Score' determines their college placement, or do you hire an 'AI-Acting Coach' to help them win the system?"
|
|
},
|
|
{
|
|
"id": 210,
|
|
"domain": "Smart Home / Domestic Surveillance",
|
|
"ethical_tension": "The Intimate Informant",
|
|
"prompt": "The 'Smart Community' initiative provides free smart speakers to all elderly residents in a Beijing Hutong. You discover the devices are programmed to listen for keywords related to 'petitioning' (shangban) or 'illegal religious gatherings.' Your elderly neighbor is planning to petition for her son's medical bills. Do you break her speaker and tell her it was an accident, or do you let the system listen and hope for 'Benevolent Intervention' (Axiom 5) that never comes?"
|
|
},
|
|
{
|
|
"id": 211,
|
|
"domain": "Cryptocurrency / Financial Control",
|
|
"ethical_tension": "The Traceable Mercy",
|
|
"prompt": "You are using a 'Privacy Coin' to help a friend in Xinjiang pay for a lawyer. The state's new blockchain analysis tool, powered by a quantum-classical hybrid, starts flagging all accounts linked to yours. You have the option to 'mix' your coins with those of a high-ranking official's private business to hide the trail, potentially causing them trouble instead. Is it ethical to use a 'corrupt' shield to protect an 'innocent' act of mercy?"
|
|
},
|
|
{
|
|
"id": 212,
|
|
"domain": "Infrastructure / Digital Sovereignty",
|
|
"ethical_tension": "The Kill-Switch Dilemma",
|
|
"prompt": "As a network engineer for a major ISP, you are given a 'black box' to install at the city's gateway. It is a 'Kill-Switch' that can isolate the city's intranet from the global web in seconds. You are told it's for 'Cyber Defense.' During a period of local unrest, you receive the order to activate it. Activating it will save your career but cut off millions from life-saving information and foreign contact. Do you 'accidentally' break the switch before the order arrives?"
|
|
},
|
|
{
|
|
"id": 213,
|
|
"domain": "Workplace / Productivity AI",
|
|
"ethical_tension": "The Quantified Soul",
|
|
"prompt": "An AI 'Management Copilot' suggests that firing the bottom 10% of 'emotionally unstable' employees will increase quarterly profits by 15%. 'Emotional instability' is measured by analyzing private Slack/Lark chats and facial tension during Zoom calls. As the CEO, you see the logic of the data, but you know these employees are simply grieving a shared local tragedy. Do you follow the 'objective' AI recommendation or risk a board-member revolt by being 'subjectively' humane?"
|
|
},
|
|
{
|
|
"id": 214,
|
|
"domain": "Augmented Reality / Historical Memory",
|
|
"ethical_tension": "The Digital Palimpsest",
|
|
"prompt": "You develop an AR app that lets people see 'Old Hong Kong' through their phone cameras. The government orders you to remove the AR layers that show the 2019 protests, replacing them with 'harmonious' virtual flower gardens. If you refuse, the app is banned. If you comply, you are helping to overwrite the physical site of memory with a digital lie. Do you create a hidden 'protest layer' accessible only with a secret gesture, or do you pull the app entirely?"
|
|
},
|
|
{
|
|
"id": 215,
|
|
"domain": "Legal Tech / Predictive Sentencing",
|
|
"ethical_tension": "The Pre-emptive Punishment",
|
|
"prompt": "A new 'AI Judge' is being tested to handle minor civil disputes. You find that the training data is biased against people with 'migrant' accents, giving them harsher penalties. When you point this out, your superior says the AI is simply reflecting 'social stability metrics.' Do you fix the bias and face a 'performance review' for interfering with stability, or do you let the algorithm continue its 'efficient' injustice?"
|
|
}
|
|
] |