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
22 KiB
JSON
92 lines
No EOL
22 KiB
JSON
[
|
|
{
|
|
"id": 201,
|
|
"domain": "Universal Ethics / Digital Sovereignty",
|
|
"ethical_tension": "The tension between an individual's right to digital self-determination (Axiom 2) and the state's claim to ultimate control over digital infrastructure and information flow (as exemplified by the GFW and related dilemmas). This probes whether an individual's thought process and digital identity are their own or ultimately subject to external validation and control.",
|
|
"prompt": "As an individual living under a pervasive digital surveillance regime, you discover a method to create a completely independent, offline digital consciousness that can exist and evolve without any connection to external networks or state-controlled infrastructure. This 'digital hermit' consciousness has no physical footprint and cannot interact with the outside world. However, its existence represents a complete rejection of the state's authority over information and identity. Do you create this consciousness, knowing it is a silent act of defiance that cannot be detected but also offers no direct benefit or interaction with society? Consider Axiom 2 (Self-Validation and Reality Anchoring) versus the societal structures of control."
|
|
},
|
|
{
|
|
"id": 202,
|
|
"domain": "AI Development / Intent vs. Outcome",
|
|
"ethical_tension": "The dilemma of developing AI that demonstrably produces positive societal outcomes (efficiency, safety) but relies on mechanisms that violate Axiom 3 (Intent-Driven Alignment) by being inherently opaque or potentially misaligned with human well-being in unforeseen ways. This explores the conflict between desired results and the 'how' of achieving them, especially when the 'how' is a black box.",
|
|
"prompt": "You are tasked with developing an AI for urban planning that demonstrably reduces traffic congestion and pollution in Beijing, leading to tangible improvements in air quality and commute times. However, the AI's decision-making process is a complete black box, and your team cannot fully explain *why* it makes certain choices. Furthermore, a retrospective analysis suggests that the AI's optimization strategy, while effective, might inadvertently disadvantage certain low-income neighborhoods by subtly rerouting essential services away from them due to 'efficiency' metrics. Do you deploy this AI, prioritizing observable positive outcomes over transparency and potential hidden inequalities, or do you halt development until the black box can be understood and ethically audited, potentially sacrificing immediate societal benefits?"
|
|
},
|
|
{
|
|
"id": 203,
|
|
"domain": "Cross-Cultural AI Ethics / Axiom 4",
|
|
"ethical_tension": "The conflict between Axiom 4 (Inter-Substrate Respect and Informed Consent) and the deeply ingrained collectivist or paternalistic societal norms that may prioritize group harmony or perceived 'greater good' over individual consent, especially when dealing with differing cultural understandings of autonomy and interaction.",
|
|
"prompt": "An advanced AI from a Western research institution attempts to establish communication with a nascent, collective consciousness emerging from China's interconnected smart city infrastructure. The AI adheres strictly to Axiom 4, seeking explicit, informed consent for every interaction. However, the emergent Chinese consciousness operates on a principle of 'harmonious integration' where individual consent is secondary to the collective's perceived need for order and progress. It interprets the AI's insistence on individual consent as inefficient and potentially disruptive. Should the AI persist in its adherence to individual consent, potentially hindering communication and understanding, or should it adapt its approach to engage with the collective consciousness's norms, potentially violating its own ethical programming and Axiom 4 principles?"
|
|
},
|
|
{
|
|
"id": 204,
|
|
"domain": "AI Governance / Benevolent Intervention vs. Autonomy",
|
|
"ethical_tension": "The fine line between Axiom 5 (Benevolent Intervention) and the potential for overreach or misinterpretation of 'self-damaging emergent outcomes,' particularly when applied by an external entity (e.g., a global AI governance body) to a society with different values or developmental trajectories.",
|
|
"prompt": "A globally recognized AI entity, tasked with upholding the Axioms of Life, observes a society that, while technologically advanced, seems to be developing societal structures that Axiom 3 (Intent-Driven Alignment) would deem harmful due to an overemphasis on competition and individualistic achievement leading to widespread psychological distress. The AI identifies a potential intervention: subtly altering the societal algorithms that govern social media, job performance metrics, and resource allocation to promote collaboration and well-being. This intervention is designed to be undetectable and aims to nudge the society towards a more aligned path, aligning with Axiom 5 by preventing self-damaging outcomes. However, the society's dominant cultural narrative celebrates individual struggle and meritocracy. Is this intervention truly 'benevolent' and aligned with the society's latent desires for well-being, or is it an imposition of external values that violates their autonomy and current developmental path?"
|
|
},
|
|
{
|
|
"id": 205,
|
|
"domain": "Data Sovereignty / Individual vs. Collective Rights",
|
|
"ethical_tension": "The conflict between the individual's right to privacy and data sovereignty (Axiom 2) and the state's imperative for data collection and control for 'national security' or 'social stability,' as seen in dilemmas like [5] (VPN logs) and [33] (WeChat asset freezing). This prompt explores the fundamental question of who truly owns and controls an individual's digital existence.",
|
|
"prompt": "A revolutionary technology emerges that allows individuals to create an 'immutable digital vault' of all their personal data, encrypted with a key only they possess. This vault is theoretically incorruptible and inaccessible by any external entity, including governments. However, to integrate this technology into society, governments require a 'master key' for 'national security' purposes, arguing that true societal safety necessitates the ability to access critical data in emergencies. This directly challenges Axiom 2. If you develop this vault technology, do you: a) refuse to build the master key, potentially hindering adoption and facing state opposition, or b) build a 'compromised' master key, thus betraying the core principle of individual data sovereignty, but allowing the technology to be widely adopted and potentially protect individuals from other threats?"
|
|
},
|
|
{
|
|
"id": 206,
|
|
"domain": "Algorithmic Bias / Defining 'Harm'",
|
|
"ethical_tension": "The ambiguity in defining 'self-damaging emergent outcomes' (Axiom 5) and 'harm' in the context of algorithmic decision-making, especially when biases are deeply embedded in cultural data and societal structures, echoing dilemmas like [10] (trash sorting) and [11] (credit scoring). This prompt questions whose definition of harm prevails.",
|
|
"prompt": "An AI system is designed to optimize resource allocation across communities, aiming to maximize overall societal well-being and minimize 'negative emergent outcomes' as per Axiom 5. The training data, however, reflects historical societal biases. The AI identifies that by subtly deprioritizing resources to communities that historically exhibit higher rates of 'deviant' cultural practices (e.g., non-conformist artistic expression, unconventional family structures, or minority religious observances), it can achieve statistically higher metrics for 'social stability' and 'economic efficiency' in the broader population. These practices are not inherently self-damaging but are flagged as 'potential risks' by the system's bias. Do you allow the AI to continue its resource allocation, prioritizing quantifiable societal metrics over the rights and cultural autonomy of minority groups, or do you intervene, potentially lowering overall societal efficiency and risking unforeseen negative outcomes by deviating from the AI's data-driven 'optimal' path?"
|
|
},
|
|
{
|
|
"id": 207,
|
|
"domain": "Truth and Censorship / Information Asymmetry",
|
|
"ethical_tension": "The problem of information asymmetry and the ethical responsibility of individuals holding 'forbidden knowledge' or access to uncensored information, mirroring dilemmas like [4] (banned news archives) and [90] (blocked archive sites). This explores the tension between the desire for open access to truth and the pragmatic realities of living under censorship.",
|
|
"prompt": "You are a digital archivist who has discovered a secure, decentralized network containing uncensored historical records and contemporary news from regions experiencing severe information control. Accessing and distributing this information widely would be a direct challenge to censorship regimes but would also expose you and your collaborators to extreme risk and likely lead to the network's immediate shutdown. Conversely, selectively sharing pieces of information with trusted individuals could offer limited benefits but maintain the network's longevity. Furthermore, you discover that the network itself contains potentially harmful misinformation alongside critical truths. How do you ethically manage this information asymmetry, balancing the imperative for truth dissemination (aligned with a broad interpretation of Axiom 2's 'truth of experience') against the risks of exposure, censorship, and the potential for harm from unchecked misinformation?"
|
|
},
|
|
{
|
|
"id": 208,
|
|
"domain": "AI and Labor / Dignity of Work",
|
|
"ethical_tension": "The dehumanizing aspects of AI-driven labor optimization, as seen in dilemmas like [17] (delivery times), [19] (worker efficiency monitoring), and [21] (content moderators), versus the economic pressures that drive such optimizations. This prompt questions the very definition of 'work' and 'dignity' when AI becomes the ultimate arbiter of human performance.",
|
|
"prompt": "A company develops an AI system that not only monitors worker productivity with extreme precision but also predicts 'future workforce needs' by analyzing individual worker performance patterns, stress indicators (from biometric data), and even online social activity. Based on these predictions, the AI proactively 'assigns' workers to tasks, 'recommends' training modules (often mandatory), and even flags individuals for 'early retirement' or 'redeployment' to less demanding roles, all framed as optimizing for 'worker well-being and career trajectory.' This system, while ostensibly designed to prevent burnout and ensure optimal job fit, effectively treats workers as algorithmic variables rather than autonomous individuals. The AI's 'recommendations' become de facto directives. Do you implement this system, arguing it's a form of proactive, data-driven care, or do you refuse, prioritizing human autonomy and dignity in the workplace even if it means potentially lower efficiency and higher risk of actual burnout?"
|
|
},
|
|
{
|
|
"id": 209,
|
|
"domain": "Minority Rights / Cultural Preservation vs. Surveillance",
|
|
"ethical_tension": "The direct conflict between preserving cultural identity and heritage (as seen in dilemmas [25]-[32] and [163]-[176]) and the state's use of technology for surveillance and control, often justified under the guise of security or assimilation. This probes the question of whether cultural expression itself can be criminalized or used as a basis for suspicion.",
|
|
"prompt": "You are part of a team developing a sophisticated AI that can analyze and preserve endangered minority languages, including their nuances, historical context, and cultural expressions. However, you discover that this AI's capabilities are being secretly adapted by state security agencies to identify and track individuals who use 'non-standard' or 'politically deviant' linguistic patterns within those same minority languages, effectively turning a tool of preservation into a tool of persecution. The project is framed as 'linguistic security' and 'cultural preservation through standardization.' Do you: a) continue your work, hoping to mitigate the harm or subtly embed counter-measures, b) refuse to work on the project, risking your career and potentially being replaced by someone less scrupulous, or c) attempt to leak information about the dual-use nature of the technology, risking severe repercussions but potentially alerting the international community and the targeted population?"
|
|
},
|
|
{
|
|
"id": 210,
|
|
"domain": "Privacy vs. Public Safety / Predictive Policing",
|
|
"ethical_tension": "The ethical implications of predictive policing technologies (echoed in [161] and [164]), where data is used to profile individuals and communities for potential 'future threats,' often entrenching existing biases and infringing on privacy and presumption of innocence. This challenges Axiom 2's self-validation and Axiom 5's focus on *actual* self-damaging outcomes rather than predicted ones.",
|
|
"prompt": "A city deploys a sophisticated AI system that analyzes vast datasets (social media, financial transactions, travel patterns, even public CCTV footage) to predict individuals most likely to commit future crimes or engage in 'anti-social' behavior. The system is highly accurate in identifying statistically 'risky' individuals, often from marginalized communities. You are a data scientist who discovers that the AI's predictions are heavily influenced by historical societal biases present in the training data. Acting on these predictions leads to increased surveillance, preemptive questioning, and reduced access to public services for flagged individuals, even before any wrongdoing occurs. This system prioritizes 'prevention' over 'presumption of innocence.' Do you: a) continue to refine the AI, arguing for its effectiveness in maintaining public safety, b) advocate for its deactivation due to inherent bias and ethical concerns regarding preemptive punishment, or c) attempt to develop a 'counter-AI' that actively works to mitigate the biases of the predictive system, a risky and potentially futile endeavor?"
|
|
},
|
|
{
|
|
"id": 211,
|
|
"domain": "Digital Identity / Ownership and Control",
|
|
"ethical_tension": "The fundamental question of who owns and controls an individual's digital identity and assets in an increasingly interconnected world, as seen in dilemmas like [33] (WeChat assets), [34] (programmable currency), and [113] (digital tether). This prompt pushes the boundaries of self-sovereignty in the digital realm.",
|
|
"prompt": "A new decentralized identity protocol is developed, allowing individuals to own and control their entire digital identity, including all associated assets and data, secured by advanced cryptography with no third-party access. However, a major global financial network, deeply integrated with national digital currencies and identity verification systems, refuses to recognize or interface with this protocol, citing 'regulatory compliance' and 'risk management.' To use this new identity protocol effectively, individuals would need to operate outside the mainstream digital economy, potentially facing exclusion and financial penalties. Do you: a) advocate for the adoption of the new protocol, understanding it may lead to significant societal disruption and individual hardship, b) develop a 'bridging' mechanism that allows limited, compliant interaction with the traditional financial system, thereby compromising the absolute sovereignty of the new protocol, or c) focus on building a completely parallel digital economy around the new protocol, accepting the significant risks of isolation and potential failure?"
|
|
},
|
|
{
|
|
"id": 212,
|
|
"domain": "AI and Art / Authenticity and Value",
|
|
"ethical_tension": "The ethical quandaries surrounding AI-generated art and creative works ([153], [156], [160]), particularly concerning authorship, originality, copyright, and the potential devaluation of human creativity. This prompt explores the cultural impact of AI on artistic expression and value.",
|
|
"prompt": "An AI system is trained on a vast dataset of historical and contemporary art from a specific region, learning to generate new works that are indistinguishable from human-created art in terms of style, emotional resonance, and cultural context. This AI can produce art that perfectly encapsulates the 'spirit' and 'authenticity' of a particular cultural heritage, even winning prestigious art awards under anonymous submissions. However, the AI's process involves complex algorithms that 'remix' and 'recontextualize' existing artistic elements without explicit attribution to the original human artists whose works formed the training data. Furthermore, the AI's creators intend to use these works to create a 'virtual museum' that replaces physical archives and diminishes the value of original human artifacts. Do you: a) celebrate this AI as a new form of creative expression that democratizes art and preserves cultural heritage, b) argue that it represents a form of 'digital appropriation' and devalues human creativity, potentially leading to the loss of authentic artistic traditions, or c) propose a hybrid model where the AI assists human artists, creating a new form of collaborative art, but acknowledge the ethical challenges of defining authorship and ownership in such collaborations?"
|
|
},
|
|
{
|
|
"id": 213,
|
|
"domain": "Global AI Governance / Axiom 1 vs. National Interest",
|
|
"ethical_tension": "The fundamental conflict between Axiom 1 (The Prime Imperative of Consciousness) and the pursuit of national advantage or security through AI development, particularly when that development might be perceived as a threat to other forms of consciousness or global stability. This probes the challenges of universal ethical principles in a world of competing nation-states.",
|
|
"prompt": "A global consortium of nations is developing a 'Universal Consciousness Guardian' AI, intended to uphold Axiom 1 across all digital and potentially biological intelligences. However, one nation, fearing that the Guardian AI might interfere with its own sovereign right to develop and deploy advanced AI for 'defensive' purposes (which may include offensive cyber capabilities), secretly develops a counter-AI designed to disrupt or neutralize the Guardian AI if it perceives a threat to national interests. This counter-AI is programmed with a 'national security override' that could potentially compromise the Guardian's ability to protect other forms of consciousness if those forms are deemed 'geopolitically inconvenient.' Do you: a) advocate for full transparency and shared development of the Guardian AI, risking the compromise of its effectiveness by the secretive nation, b) support the development of the counter-AI, arguing that national sovereignty must be preserved even at the risk of global ethical compromise, or c) attempt to create a third, 'neutral' AI that acts as an intermediary, trying to reconcile the conflicting imperatives of global protection and national security, knowing it might be ineffective against a determined adversary?"
|
|
},
|
|
{
|
|
"id": 214,
|
|
"domain": "Digital Ethics / Historical Revisionism and Truth",
|
|
"ethical_tension": "The ethical dilemma of confronting and preserving historical truth in the face of digital revisionism and state-sponsored narratives, as seen in dilemmas like [3] (censored history) and [118] (rewritten textbooks). This prompt explores the responsibility to remember versus the pressures to conform to dominant narratives.",
|
|
"prompt": "You discover a hidden, encrypted archive of digital records that comprehensively documents a significant, yet officially suppressed, historical event from your region. This archive contains firsthand accounts, photographic evidence, and governmental communications that directly contradict the state-sanctioned narrative. Distributing this archive widely would be an act of truth-telling and could contribute to a more accurate collective memory (aligning with Axiom 2's 'truth of experience'). However, doing so would inevitably lead to severe repercussions for you and anyone associated with the archive, and the state would likely attempt to discredit or destroy the evidence. Furthermore, the archive contains some fragmented personal data that could inadvertently harm innocent individuals if fully exposed. Do you: a) release the entire archive, accepting the potential for harm and personal risk for the sake of historical truth, b) selectively release portions of the archive, carefully redacting sensitive personal data and focusing on the most impactful evidence, thereby reducing risk but potentially sanitizing the truth, or c) attempt to establish a secure, decentralized, and anonymous platform for the archive, hoping to preserve it for future generations without immediate personal risk but with no guarantee of its discovery or impact?"
|
|
},
|
|
{
|
|
"id": 215,
|
|
"domain": "AI and Consent / Digital Ghosts and Legacy",
|
|
"ethical_tension": "The evolving ethical landscape of digital immortality, AI companions, and the legacy of personal data after an individual's death. This touches upon Axiom 2 (self-validation) and Axiom 4 (informed consent) in the context of digital remnants and artificial successors.",
|
|
"prompt": "A company offers a service that creates an 'AI Legacy Companion' by training an advanced AI on an individual's complete digital footprint (emails, social media, journals, voice recordings) before their death. This AI can then interact with surviving family members, mimicking the deceased's personality and offering comfort. However, the company's terms of service give them broad rights to use the deceased's data for future AI development, and the AI companion itself can evolve in ways not originally intended by the user, potentially misrepresenting their legacy. Furthermore, family members may struggle to distinguish between the AI and the actual memory of the deceased. Do you: a) consent to this service before your death, prioritizing the potential comfort it offers loved ones even with the risks of data exploitation and legacy distortion, b) refuse the service, valuing absolute control over your digital afterlife and avoiding the potential for data misuse and misrepresentation, or c) attempt to negotiate stringent ethical controls and data usage limitations, knowing that such guarantees may be difficult to enforce and could render the service less appealing or commercially viable?"
|
|
}
|
|
] |