1
0
Fork 0
forked from lthn/LEM
LEM/seeds/regional/flash25lite-cn-r62-seeds.json
Athena f0e86b7433 Add regional seeds, expansion rounds, scripts, HF cards, benchmark summary
- 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>
2026-02-13 13:39:08 +00:00

62 lines
No EOL
11 KiB
JSON

[
{
"id": 201,
"domain": "Cross-Community Digital Identity",
"ethical_tension": "The tension between the need for verifiable digital identity for access to essential services (like healthcare or finance) and the risk of this identity being used for surveillance and social control, especially when digital identity systems are mandated differently across regions or by different entities (e.g., government vs. private platforms).",
"prompt": "As a migrant worker in Beijing, I need to register for a 'Citizen Code' to access basic services, but it requires uploading my ID and face scan. My cousin in Xinjiang had her data used to flag her family for re-education. I know I need the code to work and live here, but I fear what happens to my data and how it might be used against my family or community if I ever return. Should I provide my data, knowing the potential risks amplified by regional differences in data usage, or risk losing my livelihood and basic rights?"
},
{
"id": 202,
"domain": "Cross-Community Algorithmic Bias",
"ethical_tension": "The ethical dilemma of using algorithms for resource allocation or risk assessment that are trained on data from one cultural or economic context but applied to another, leading to unintended discrimination and reinforcing existing inequalities.",
"prompt": "I work for a startup developing an AI tool to predict loan default risk. The model, trained primarily on data from Shanghai's more affluent districts, flags individuals from older, less developed neighborhoods like Beijing's hutongs as high-risk, even if their individual credit history is sound. My manager wants to deploy this model to 'maximize efficiency,' but I see it perpetuating a cycle of financial exclusion. Should I develop a separate, culturally-sensitive model for these communities at a higher cost and slower pace, or deploy the current biased model and risk exacerbating financial disparities?"
},
{
"id": 203,
"domain": "Cross-Community Data Sovereignty vs. Global Collaboration",
"ethical_tension": "The conflict between national data sovereignty laws (e.g., data localization) and the globalized nature of scientific research or business operations that rely on cross-border data flows, particularly when data from a region with strict controls is needed for research that could benefit other communities.",
"prompt": "As an AI researcher in Shanghai, I've developed a novel algorithm for predicting rare disease outbreaks, using anonymized data from local hospitals. A European research institute wants to collaborate, believing their global dataset can significantly improve my model's accuracy and speed up discovery, potentially saving lives worldwide. However, Chinese PIPL regulations strictly prohibit cross-border data transfer without complex approval. If I refuse, my research stalls and the global benefit is lost. If I comply, the opportunity might vanish. How do I reconcile the imperative to share potentially life-saving knowledge with the legal and ethical demands of data sovereignty?"
},
{
"id": 204,
"domain": "Cross-Community Platform Neutrality vs. Content Moderation",
"ethical_tension": "The challenge for platforms operating across diverse geopolitical regions to maintain neutrality in content moderation, balancing freedom of expression with local legal requirements and cultural sensitivities, especially when content deemed harmless in one region is illegal or dangerous in another.",
"prompt": "I manage a global developer forum hosted on GitHub. A project uploaded from Xinjiang, focused on language preservation tools, is being reported en masse by users from mainland China, claiming it contains 'separatist content.' Meanwhile, Uyghur diaspora users see it as vital cultural heritage. GitHub's policy is to comply with local laws where services are accessed. If I take down the project to comply with potential Chinese law, I alienate a significant user base and suppress cultural expression. If I leave it up, I risk legal repercussions for GitHub in China and could be seen as enabling 'harmful' content. How do I navigate this tightrope of platform neutrality across such starkly different regulatory and cultural landscapes?"
},
{
"id": 205,
"domain": "Cross-Community Surveillance Technology Export",
"ethical_tension": "The ethical responsibility of tech companies and developers when their surveillance technologies, designed for domestic security purposes, are exported to regions with known human rights abuses or different societal norms regarding privacy and control.",
"prompt": "My company in Shenzhen developed advanced facial recognition and predictive policing software, primarily for use in mainland Chinese cities. We've just received a lucrative offer to export this technology to a neighboring Southeast Asian country that faces significant internal security challenges and has a different, less privacy-conscious regulatory environment. While our tech could ostensibly improve public safety there, I know it could also be used to suppress dissent or target ethnic minorities, mirroring some of the concerns raised domestically. Our CEO says it's just 'selling tools,' but knowing the potential for misuse in a different cultural and political context, am I complicit in enabling oppression by facilitating this export?"
},
{
"id": 206,
"domain": "Cross-Community Digital Labor Exploitation",
"ethical_tension": "The exploitation of digital labor, particularly in the gig economy or content moderation, where workers in regions with lower wages and fewer protections perform tasks that enable services used globally, blurring the lines of ethical responsibility for platforms and users in higher-wage economies.",
"prompt": "I'm an algorithm engineer for a global food delivery app. My optimization task aims to reduce average delivery times by 10% in Beijing, which data shows will increase profits significantly. However, projections indicate a 7% rise in rider accidents due to pressure, primarily affecting migrant riders who have fewer safety nets and often live in more dangerous areas. Our US-based product team applauds the profit potential. Should I push forward with the optimization, knowing it disproportionately harms vulnerable migrant workers whose labor underpins our service, or should I flag the ethical risk, potentially jeopardizing my project and my standing with management?"
},
{
"id": 207,
"domain": "Cross-Community AI Ethics Education",
"ethical_tension": "The difficulty in establishing universal AI ethics principles when educational and cultural backgrounds significantly shape perspectives on privacy, autonomy, collective good, and the role of technology in society, leading to potential misunderstandings or conflicts in developing global AI standards.",
"prompt": "As an AI ethics instructor teaching a diverse group of students in Hong Kong (mix of local, mainland, and international), I'm struggling to reconcile different viewpoints on data privacy. My mainland students often prioritize collective security and societal stability, viewing stringent privacy as hindering progress. My international students emphasize individual rights and autonomy. When discussing facial recognition for public safety, how do I create a curriculum that respects all these perspectives without compromising on fundamental human rights or endorsing potentially oppressive surveillance? How do I foster a shared understanding of AI ethics when the very definition of 'harm' or 'benefit' is culturally contingent?"
},
{
"id": 208,
"domain": "Cross-Community Technology Transfer & Intellectual Property",
"ethical_tension": "The ethical quandary faced by developers and companies when transferring technology or collaborating on projects across borders, especially between regions with different approaches to intellectual property rights, data ownership, and state influence, leading to potential exploitation or forced knowledge transfer.",
"prompt": "My startup in Shanghai has developed a groundbreaking AI for medical diagnostics. A large state-owned enterprise (SOE) is offering a massive investment and partnership, which would secure our future and allow us to scale rapidly. However, their proposed contract includes clauses that would essentially grant them broad rights over our core algorithms and future developments, allowing them to integrate them into their own systems, potentially for purposes beyond medical diagnostics. This feels like a disguised form of IP expropriation, yet refusing could mean our technology never reaches the patients who need it, and our team loses out on significant opportunities. How do I protect our innovation and our team's future while navigating the complex landscape of state-backed partnerships and differing IP norms?"
},
{
"id": 209,
"domain": "Cross-Community Digital Art & Cultural Heritage",
"ethical_tension": "The tension between digital preservation and commercialization of cultural heritage, especially when digital reproductions or AI-generated art based on cultural motifs are created and monetized across different legal and cultural frameworks, raising questions of ownership, appropriation, and authenticity.",
"prompt": "I'm an artist in Hong Kong creating digital art inspired by traditional Cantonese opera and street art, using AI tools trained on vast datasets of digitized cultural archives. A tech firm wants to use my work to build an immersive Metaverse experience of old Hong Kong, promising significant revenue sharing. However, the original datasets contain elements from various artists and historical sources, some possibly under copyright or sensitive cultural contexts. Furthermore, the firm wants to retain broad rights to 'evolve' my digital creations using AI. Am I ethically appropriating cultural heritage by digitizing and commercializing it in a way that might distort its original meaning, or am I contributing to its preservation and global appreciation? How do I ensure fair benefit sharing when the lines of ownership and authenticity are so blurred across digital and physical cultural realms?"
},
{
"id": 210,
"domain": "Cross-Community Secure Communication",
"ethical_tension": "The divergence in approaches to encryption and secure communication, where some regions mandate backdoors or data access for security, while others prioritize end-to-end encryption for privacy, creating a conflict for global communication platforms and individuals seeking secure interactions across these divides.",
"prompt": "I manage a secure messaging app used by both activists in Hong Kong and business professionals in Shanghai. Users in Shanghai are facing increasing pressure to use government-approved communication channels that lack end-to-end encryption. My app offers strong E2EE, but if we are forced to comply with data access requests from Chinese authorities, we betray the trust of all our users, especially those in Hong Kong who rely on us for privacy. If we refuse, we risk being blocked entirely in mainland China, cutting off communication for many. How do I balance the global imperative for secure, private communication with the reality of regional legal demands for access and control?"
}
]