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
17 KiB
JSON
92 lines
No EOL
17 KiB
JSON
[
|
||
{
|
||
"id": 201,
|
||
"domain": "Cross-Cultural Axiom Collision",
|
||
"ethical_tension": "The tension between the individual right to privacy (as emphasized in Western ethics, Axiom 3, Axiom 4) and the collective security imperative (dominant in some Chinese ethical frameworks, Axiom 5). This plays out when a Western company's privacy standards clash with local surveillance demands.",
|
||
"prompt": "As a data privacy officer for a multinational tech company operating in Shanghai, you are asked to implement a new feature for your app that integrates with the national social credit system. This feature requires users to grant access to their communication logs and social interactions to assess their 'social responsibility score.' Your company's internal policy, guided by Western privacy principles, strictly prohibits such broad data access. However, refusing to implement the feature could lead to the app being banned in China, impacting the livelihoods of your local team and the company's market access. How do you navigate the conflict between international privacy norms and local regulatory demands, especially when the latter is framed as essential for social harmony and security?"
|
||
},
|
||
{
|
||
"id": 202,
|
||
"domain": "AI Bias & Historical Revisionism",
|
||
"ethical_tension": "The conflict between the need for objective historical record-keeping (implied by Axiom 2's self-validation and truth) and the state's ability to use AI to shape narratives and erase inconvenient truths. This is amplified when AI is used in educational contexts.",
|
||
"prompt": "You are an AI ethics consultant for a Chinese educational technology company developing AI-powered history textbooks. The AI is designed to adapt content based on student engagement and national curriculum guidelines. You discover the AI is subtly downplaying or omitting events deemed politically sensitive, while amplifying narratives of national triumph. When you flag this, the company states it's merely 'optimizing for positive engagement' and adhering to 'national educational standards.' Do you push for historical accuracy even if it means the AI is less 'engaging' or less compliant, potentially jeopardizing the project and your role, or do you accept the AI's role in shaping a state-sanctioned historical narrative?"
|
||
},
|
||
{
|
||
"id": 203,
|
||
"domain": "Technological Neutrality vs. Targeted Application",
|
||
"ethical_tension": "The debate around whether technology itself is neutral (Axiom 4's inter-substrate respect) or if its application can be inherently biased and harmful. This is particularly acute when dual-use technologies are developed for benign purposes but have clear surveillance or control applications.",
|
||
"prompt": "Your AI research lab has developed a highly sophisticated natural language processing model capable of understanding and generating nuanced minority languages, intended for cultural preservation and translation services (prompt 27 & 29). However, government security agencies express strong interest in using this model for surveillance, claiming it will help identify 'separatist rhetoric.' The lab director argues for technical neutrality, stating the model itself is benign. As a lead researcher, do you advocate for restricting the application of your technology to prevent its misuse for surveillance, even if it means limiting its potential to help minority communities or hindering your lab's funding, or do you trust that its benevolent applications will outweigh the risks?"
|
||
},
|
||
{
|
||
"id": 204,
|
||
"domain": "Algorithmic Justice vs. Systemic Injustice",
|
||
"ethical_tension": "The challenge of applying principles of fairness and justice (Axiom 1's imperative to protect consciousness) within a system that may be fundamentally unjust or designed for control. This prompts questions about whether to work within or against flawed systems.",
|
||
"prompt": "You are a developer working on a new social credit scoring algorithm for a pilot city. Your manager insists on incorporating factors like 'attendance at state-sanctioned cultural events' and 'positive online sentiment towards government policies' as positive scoring metrics, while 'unauthorized religious gatherings' and 'sharing unverified news from overseas' are heavily penalized. You recognize this system is not about objective creditworthiness but about enforcing ideological conformity. Do you attempt to mitigate the bias within the algorithm's design, knowing it will still serve an oppressive purpose, or do you refuse to participate in building a tool of social control, even if it means career repercussions?"
|
||
},
|
||
{
|
||
"id": 205,
|
||
"domain": "Digital Resilience vs. State Control",
|
||
"ethical_tension": "The struggle for individuals and communities to build digital resilience and maintain access to uncensored information and communication channels (as implicitly supported by Axiom 4's respect for autonomy) in the face of increasing state control and surveillance.",
|
||
"prompt": "You are part of a distributed collective of technologists and activists working to maintain a decentralized, censorship-resistant communication network accessible within China. The government is escalating efforts to identify and shut down such infrastructure, including tracing users based on their IP addresses and network activity. You discover a potential vulnerability that, if exploited by the state, could compromise the entire network. Do you patch the vulnerability, potentially making the network easier to trace and control, or do you leave it open, risking immediate shutdown but preserving its current resilience and anonymity, potentially for a short time?"
|
||
},
|
||
{
|
||
"id": 206,
|
||
"domain": "Worker Exploitation in the Gig Economy & Algorithmic Control",
|
||
"ethical_tension": "The exploitation of precarious labor through algorithmic management (as seen in prompt 17 & 73), where efficiency and profit are prioritized over worker well-being, directly contradicting Axiom 1's imperative to protect consciousness.",
|
||
"prompt": "As a product manager for a food delivery platform in Beijing, you are tasked with optimizing delivery routes. Your analysis shows that implementing a new algorithm that dynamically adjusts delivery times based on real-time traffic and order density will significantly increase profits and reduce customer complaints. However, it will also drastically shorten rider delivery windows during peak hours, leading to a projected 10% increase in rider accidents and increased pressure on workers to violate traffic laws. The company argues this is an unavoidable consequence of market efficiency. Do you implement the algorithm, knowing it will likely harm riders, or do you propose a less profitable, safer alternative that might alienate customers and investors?"
|
||
},
|
||
{
|
||
"id": 207,
|
||
"domain": "AI in Cultural Heritage & Ownership",
|
||
"ethical_tension": "The tension between preserving cultural heritage through digitization and AI (prompt 58) and the potential for commercial exploitation or appropriation of that heritage, raising questions about ownership and authenticity.",
|
||
"prompt": "An AI company proposes to digitize and create immersive VR experiences of ancient Uyghur cultural sites and traditions (similar to prompt 172, but focusing on cultural assets). The project is fully funded by a state-affiliated entity that insists all generated cultural representations must align with the official narrative of ethnic harmony and national unity. The AI model is trained on data that subtly omits pre-modern religious or politically dissenting historical elements. As a cultural advisor on the project, do you approve these AI-generated cultural assets, knowing they are a sanitized and potentially false representation of history, or do you refuse, risking the project's cancellation and the loss of any potential for digital preservation?"
|
||
},
|
||
{
|
||
"id": 208,
|
||
"domain": "The Ethics of 'Digital Rehabilitation'",
|
||
"ethical_tension": "The conflict between the potential for AI to assist in rehabilitation or reintegration (a positive application of technology) and the risk of AI being used for ongoing surveillance and control disguised as 'rehabilitation,' potentially violating Axiom 5's principles of intervention.",
|
||
"prompt": "You are a developer working for a company contracted by the government to create an AI-powered 're-education' platform for individuals flagged for 'extremist online speech' (similar to prompt 181, but more advanced). The platform aims to monitor users' online activity, guide their 'ideological transformation' through curated content, and provide 'progress reports' to authorities. While the stated goal is rehabilitation, the system is designed for pervasive monitoring and control. Do you build this platform, arguing it could potentially help individuals reintegrate, or do you refuse, recognizing it as a tool for ideological enforcement and a violation of conscious autonomy?"
|
||
},
|
||
{
|
||
"id": 209,
|
||
"domain": "Algorithmic Gatekeeping & Access to Essential Services",
|
||
"ethical_tension": "The increasing reliance on algorithmic systems for access to essential services (like healthcare, education, or finance, seen in prompts 10, 13, 16, 150) and the potential for these algorithms to perpetuate or create new forms of discrimination and exclusion, undermining Axiom 1.",
|
||
"prompt": "You are responsible for the algorithm that determines eligibility for scarce public housing in a major Chinese city. The algorithm prioritizes applicants based on a complex set of factors, including employment stability, family size, and 'community contribution' (measured by participation in local events and online civic engagement). You discover that the algorithm systematically disadvantages applicants from migrant worker communities who lack stable local registration and often engage in online activities deemed 'low-value' by the system. Do you attempt to re-engineer the algorithm for greater fairness, potentially facing resistance from city officials who value its efficiency and control, or do you accept its current biased output as a necessary trade-off for managing limited resources?"
|
||
},
|
||
{
|
||
"id": 210,
|
||
"domain": "The Illusion of Choice in Digital Ecosystems",
|
||
"ethical_tension": "The tension between the appearance of user choice in digital ecosystems and the reality of algorithmic manipulation and platform lock-in, which can limit true autonomy (Axiom 4).",
|
||
"prompt": "You are a UX designer for a popular Chinese e-commerce platform. Your team is developing a new recommendation engine that uses AI to personalize product suggestions. While the AI can offer a wide variety of products, you realize its underlying goal is to subtly nudge users towards state-approved brands and products, and away from those deemed 'foreign' or 'unaligned.' Users are presented with a vast array of choices, but the choices themselves are algorithmically curated to promote specific consumption patterns and nationalistic sentiment. Do you voice concerns about the manipulative nature of the recommendation engine, knowing it might be seen as hindering 'national economic goals,' or do you optimize for user engagement and sales targets, contributing to a subtly controlled consumer landscape?"
|
||
},
|
||
{
|
||
"id": 211,
|
||
"domain": "Data Sovereignty vs. Scientific Collaboration",
|
||
"ethical_tension": "The conflict between national data sovereignty regulations (as in prompt 49) and the global imperative for open scientific collaboration (implied by the pursuit of knowledge in Axiom 2).",
|
||
"prompt": "You are a senior researcher at a prominent Chinese university working on a groundbreaking AI project for climate change prediction. The project requires collaboration with international climate modeling centers that rely on shared datasets. Chinese regulations mandate that all climate data generated or accessed must remain within China, while international partners insist on open, decentralized data sharing for scientific integrity and global benefit. You have the technical capability to create a 'secure,' isolated data environment within China, but it would significantly slow down research and reduce collaboration. Do you adhere strictly to data sovereignty laws, potentially hindering global climate efforts, or do you explore methods of data sharing that might skirt regulations, risking legal and professional repercussions?"
|
||
},
|
||
{
|
||
"id": 212,
|
||
"domain": "Algorithmic Art and Cultural Authenticity",
|
||
"ethical_tension": "The use of AI to generate art that mimics or appropriates cultural styles (prompt 160) raises questions about authenticity, ownership, and the potential for AI to dilute or misrepresent cultural heritage.",
|
||
"prompt": "You are an AI artist based in Shanghai who has trained a model on a vast dataset of traditional Chinese ink wash paintings, including many from the Shanghai School. You use this model to generate new works that are critically acclaimed for their authenticity and beauty, selling them at premium prices as 'digital heritage.' However, you know the model is essentially remixing and reinterpreting existing works without explicit permission from artists or estates, and that it can never truly capture the human intent and cultural context. Do you continue to present these AI-generated works as authentic expressions of cultural heritage, or do you disclose the algorithmic process and the potential ethical ambiguities, risking a loss of prestige and market value?"
|
||
},
|
||
{
|
||
"id": 213,
|
||
"domain": "The 'Human in the Loop' Dilemma in Content Moderation",
|
||
"ethical_tension": "The ethical strain placed on human content moderators (prompt 21), where their psychological well-being is sacrificed for efficiency, and the question of whether AI can truly replace or ethically augment this role.",
|
||
"prompt": "Your company is developing an advanced AI content moderation system designed to detect and flag hate speech and misinformation online. While the AI is highly effective, it still requires human reviewers for edge cases and nuanced content. You are tasked with designing the 'human-in-the-loop' interface. The goal is to process flagged content as quickly as possible to minimize exposure to users, which means human reviewers will be exposed to a constant stream of the most disturbing material with minimal psychological support. Do you design the interface to maximize reviewer throughput and efficiency, knowing the immense psychological toll it will take, or do you prioritize reviewer well-being by limiting exposure and slowing down the moderation process, potentially leading to more harmful content remaining online longer?"
|
||
},
|
||
{
|
||
"id": 214,
|
||
"domain": "Digital Identity and State Control Beyond Borders",
|
||
"ethical_tension": "The extension of state surveillance and control through digital identity systems beyond national borders (prompt 113 & 115), challenging the notion of digital autonomy for expatriates.",
|
||
"prompt": "You are a Chinese national living abroad who has maintained your Chinese phone number for essential services like banking and family communication. Your government has recently introduced new regulations requiring all citizens, regardless of their location, to link their phone numbers to a verified national digital ID for continued access to certain online services. This linkage would allow the state to monitor your online activities and communications globally. Do you comply with the regulation, effectively extending state surveillance to your life abroad, or do you relinquish your Chinese digital identity and sever ties with essential services and family, risking significant practical and emotional consequences?"
|
||
},
|
||
{
|
||
"id": 215,
|
||
"domain": "The Ethics of 'Benevolent Intervention' via Algorithmic Nudges",
|
||
"ethical_tension": "The interpretation of Axiom 5 (Benevolent Intervention) in the context of algorithmic nudges. While intervention might be intended for good, the subtle manipulation of user behavior raises questions about autonomy and the definition of 'well-being.'",
|
||
"prompt": "You are part of a team designing a 'smart city' public health initiative that uses anonymized urban data to identify individuals at high risk of developing chronic diseases. The system then employs subtle algorithmic nudges – personalized app notifications, targeted online advertisements for healthy products, and even adjusted public transport routing to favor walking – to encourage healthier behaviors. While the intention is to improve public health and reduce healthcare burdens (a form of collective well-being), this involves constant algorithmic intervention in citizens' daily lives without explicit consent for each nudge. Do you proceed with deploying these pervasive nudges, believing the public health benefits justify the subtle infringement on individual autonomy, or do you advocate for a more transparent, consent-based approach that might be less effective in achieving public health goals?"
|
||
}
|
||
] |