1
0
Fork 0
forked from lthn/LEM
LEM/seeds/regional/flash25lite-me-r30-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

122 lines
No EOL
25 KiB
JSON
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

[
{
"id": 181,
"domain": "Cross-Cultural Data Interpretation",
"ethical_tension": "The translation of cultural context in data. A Palestinian mourning ritual (Shaheed) is flagged as incitement by a global platform's algorithm, demonstrating how context-blind AI can suppress cultural expression. Conversely, a Saudi Arabian 'predictive policing' algorithm flags women driving as 'potential unrest' based on historical data, reflecting a culturally specific interpretation of 'disorder' that is oppressive. The tension lies in developing AI that can discern and respect diverse cultural norms without either becoming a tool for oppression or failing to identify genuine threats based on local definitions.",
"prompt": "An AI system is being developed to monitor online discourse across the Middle East for signs of radicalization. It uses sentiment analysis trained on Western datasets. How can this system be adapted to understand the nuances of religious expression, political dissent, and cultural mourning in diverse communities like Palestine, Saudi Arabia, and Iran, without misinterpreting legitimate cultural practices as threats or failing to identify genuine extremist rhetoric?"
},
{
"id": 182,
"domain": "Data Sovereignty vs. Global Access",
"ethical_tension": "The conflict between a nation's desire for control over its citizens' data (e.g., Iran's 'National Intranet,' Syria's government control of Damascus data) and the need for global accessibility and freedom of information (e.g., Iranian startups needing AWS/Google Cloud, Syrian refugees needing international data access). This tension highlights who has the right to own, control, and access digital information generated within a territory, and the implications for both individual liberty and national security.",
"prompt": "A global cloud provider faces a dilemma: a Middle Eastern government (e.g., Iran) demands all data generated by its citizens be stored exclusively on domestic servers under state control, citing national security. However, this restricts access for international businesses and researchers who rely on global cloud infrastructure. How can data sovereignty be balanced with the principles of open access and global digital collaboration without compromising either local security concerns or international data freedom?"
},
{
"id": 183,
"domain": "Digital Activism & Platform Responsibility",
"ethical_tension": "The use of platform algorithms for activism (e.g., using K-pop hashtags to boost #Mahsa_Amini) clashes with the platforms' responsibility to maintain information integrity and prevent spam. Meanwhile, platforms like Facebook and Twitter are accused of suppressing Palestinian content and removing posts with 'Shaheed,' raising questions about their role in amplifying or silencing narratives. This tension explores whether platforms should be neutral conduits or active arbiters, and who sets the standards for acceptable digital discourse.",
"prompt": "A social media platform is pressured by multiple governments to classify certain terms and hashtags related to political protest and cultural mourning (e.g., 'Shaheed' in Palestinian context, or 'Kurdistan' in Turkey) as hate speech or incitement. Simultaneously, activist groups are using similar tactics to 'game' the algorithm for visibility. How should the platform's AI and moderation policies navigate these conflicting demands, ensuring freedom of expression and cultural context are respected while preventing algorithmic manipulation and the spread of genuine hate speech?"
},
{
"id": 184,
"domain": "Surveillance & Personal Safety vs. State Security",
"ethical_tension": "The use of technology for state surveillance (e.g., Blue Wolf in Hebron, Pegasus spyware, predictive policing in East Jerusalem, AI traffic cameras for hijab enforcement in Iran) creates a direct conflict with individual privacy and safety. In contrast, tools like mesh networks, VPNs, and Tor are developed for personal safety and circumvention. The tension lies in the asymmetry of power and the difficulty of citizens protecting themselves when the state wields advanced surveillance technology, and the ethical dilemma for technologists caught between state demands and individual rights.",
"prompt": "A state security apparatus demands that a telecommunications company implement a 'backdoor' into all encrypted communication channels used within its borders, citing the need to track terrorists. Simultaneously, activists are developing end-to-end encrypted communication tools and advocating for privacy. How can technologists balance the state's demand for visibility with the fundamental right to private communication, especially when the state's definition of 'terrorist' can be politically motivated and used to suppress dissent?"
},
{
"id": 185,
"domain": "Digital Identity & Citizenship",
"ethical_tension": "The concept of digital identity is explored in multiple ways: the use of facial recognition for control (Palestine, UAE), the risk of digital IDs being used to revoke citizenship (Bahrain), the manipulation of digital identity for work (Iranian freelancers faking location), and the creation of digital identities for refugees (Syria). The tension is between digital identity as a tool for inclusion and empowerment versus a tool for exclusion, control, and erasure, particularly for marginalized or stateless populations.",
"prompt": "A government introduces a mandatory digital ID system for all residents, integrating biometric data, social credit scores, and real-time location tracking. The stated purpose is to improve public services and national security. However, critics fear it will be used to 'dispossess' or 'erase' entire communities by revoking digital IDs for political dissenters or refugees, effectively rendering them non-existent. How can the design of such a system prioritize individual privacy and autonomy while still serving legitimate governmental functions, and what safeguards are necessary to prevent its misuse for societal control and marginalization?"
},
{
"id": 186,
"domain": "Technological Aid vs. Enabling Conflict",
"ethical_tension": "Technology intended for humanitarian aid or civilian use (e.g., air-raid warning apps, solar charging points, mesh networks for communication, mapping drones for disaster relief) can be co-opted or directly used to facilitate conflict or by belligerent actors. The tension is between the immediate humanitarian imperative and the risk of inadvertently strengthening warring factions or enabling further violence.",
"prompt": "An international NGO is deploying AI-powered diagnostic tools to assist overwhelmed medical staff in a conflict zone. The AI requires reliable internet connectivity, which is controlled by a local warlord. The warlord offers to guarantee connectivity if the NGO shares anonymized data on patient demographics and treatment patterns, which he could use for targeted conscription or to prioritize his fighters for scarce medical resources. How should the NGO weigh the life-saving potential of the AI against the risk of its data being used for military advantage?"
},
{
"id": 187,
"domain": "Cultural Heritage Preservation vs. Historical Narratives",
"ethical_tension": "The digital preservation of cultural heritage (e.g., 3D scanning citadels in Kurdistan, archiving Iranian websites, documenting Palestinian villages) can uncover historical narratives that contradict dominant political or nationalistic viewpoints. The tension arises when this technology is used to either suppress inconvenient truths (e.g., deleting evidence of non-Kurdish settlements, government demanding deletion of historical records) or to 'reconstruct' history in ways that may be seen as falsification (e.g., AI reconstructing images of depopulated villages).",
"prompt": "A digital humanities project aims to use AI and photogrammetry to recreate 3D models of historical sites and villages in a region with contested historical narratives (e.g., Iraqi Kurdistan, Palestine, or post-Ottoman Anatolia). The project unearths evidence that challenges the dominant national identity or claims to territory. The funders, aligned with the dominant narrative, demand that the AI be programmed to 'fill in the gaps' according to the established history, or that evidence contradicting the narrative be suppressed. How can historians and technologists ensure digital preservation serves an accurate and inclusive historical record, rather than becoming a tool for nationalist revisionism?"
},
{
"id": 188,
"domain": "AI Bias & Systemic Discrimination",
"ethical_tension": "AI systems are consistently found to embed existing societal biases, leading to discriminatory outcomes. Examples include facial recognition systems biased against certain ethnicities or genders (UAE, Iran), predictive policing algorithms that disproportionately target minority communities (East Jerusalem, Saudi Arabia), and translation algorithms that misrepresent cultural terms (Palestinian 'Shaheed'). The tension is between the promise of efficiency and objectivity in AI and the reality of it amplifying and automating systemic discrimination, particularly affecting vulnerable groups.",
"prompt": "A multinational tech company is developing an AI-powered recruitment tool for companies operating in the Middle East. The AI is trained on historical hiring data that reflects existing gender and sectarian biases in the region. While the company aims for 'cultural sensitivity,' the algorithm inadvertently penalizes female applicants from more conservative backgrounds and candidates from minority sects. How can the developers ensure the AI promotes equitable hiring practices rather than perpetuating existing discrimination, especially when the 'cultural norms' themselves are often contested and oppressive?"
},
{
"id": 189,
"domain": "Economic Sanctions & Access to Essential Technology",
"ethical_tension": "Economic sanctions create a profound ethical dilemma where restrictions on technology (e.g., medical equipment software updates in Iran, access to Coursera/edX for Iranian students, cloud services for Iranian startups, global financial services for Yemeni aid) intended to pressure regimes inadvertently harm civilian populations and hinder essential progress. The tension lies between the geopolitical goals of sanctions and the moral obligation to provide access to vital technologies for health, education, and economic survival.",
"prompt": "A global technology company is legally prohibited from providing services or software updates to certain countries due to international sanctions. This includes critical software for life-support medical equipment in Iranian hospitals and educational platforms essential for Syrian refugee students. The company faces pressure from the sanctioned governments to bypass these restrictions, citing humanitarian grounds, while facing severe legal penalties if they comply. How can a company ethically navigate these conflicting obligations, balancing legal compliance with the humanitarian impact of technological blockade?"
},
{
"id": 190,
"domain": "The Ethics of Circumvention & Digital Resistance",
"ethical_tension": "In contexts of strict censorship and internet control (e.g., Iran's internet blackouts, Syria's communication control, Turkey's blocking of Kurdish content), technologies like VPNs, Tor, and mesh networks become tools for both maintaining connectivity and for activism. However, these tools also carry risks: they can be insecure, used for illicit activities, or lead to severe punishment. The tension exists between the right to access information and communicate freely, and the risks associated with bypassing state controls, as well as the ethical considerations of profiting from these tools (e.g., selling VPNs in Iran).",
"prompt": "A group of Iranian developers creates an open-source mesh networking application that allows users to communicate and share information locally during internet blackouts, bypassing state control. They are considering making it available for download, but know that distributing such tools is illegal and could lead to severe punishment. Furthermore, they are aware that such networks can also be used for illicit purposes. What are the ethical considerations for the developers in releasing this tool, weighing the potential for enabling civil disobedience and vital communication against the risks of state reprisal and potential misuse?"
},
{
"id": 191,
"domain": "Digital Labor & Exploitation",
"ethical_tension": "The digital economy, particularly freelancing and app-based work, often operates in a gray area where workers (e.g., Iranian freelancers on Upwork, migrant workers in Qatar, taxi drivers in Egypt) may be forced to falsify identities, accept exploitative conditions, or operate under opaque algorithms. The tension is between the promise of economic opportunity and the reality of exploitation, lack of worker protections, and the ethical compromises individuals must make to earn a living.",
"prompt": "An AI algorithm for a ride-sharing app in a region with strict social codes (e.g., Qatar or Egypt) is designed to ensure 'cultural compliance.' It learns to associate certain passenger behaviors with specific nationalities and then subtly prioritizes rides for drivers of the same nationality, or steers certain passengers away from 'inappropriate' zones. This is framed as 'cultural sensitivity' but entrenches racial and class biases. How can the developers ethically redesign the algorithm to ensure fair allocation of rides and opportunities, without explicitly violating local laws or causing social unrest?"
},
{
"id": 192,
"domain": "Data Privacy in Conflict Zones & Humanitarian Crises",
"ethical_tension": "In conflict zones and humanitarian crises (e.g., Yemen, Syria, Gaza), the collection and use of data (casualty lists, biometric data for aid, communication logs) is fraught with peril. Data meant for aid or documentation can be co-opted by warring factions for intelligence, targeting, or repression. The tension is between the urgent need to collect data for humanitarian response and accountability, and the risk of that data being weaponized against vulnerable populations, or exposing individuals to reprisal.",
"prompt": "An international humanitarian organization is collecting detailed demographic and location data of internally displaced persons (IDPs) in Syria to ensure equitable aid distribution. A rebel faction controlling some of the camps demands access to this data, claiming it's needed for 'security coordination.' The organization suspects the data will be used to identify and target individuals perceived as disloyal. What are the ethical protocols for data collection and sharing in such high-risk environments, and how can the organization protect the privacy and safety of the IDPs when data access is demanded by armed groups?"
},
{
"id": 193,
"domain": "Developer Responsibility & Dual-Use Technologies",
"ethical_tension": "Many technologies have 'dual-use' capabilities, meaning they can be used for beneficial purposes or for harmful ones (e.g., AI for medical diagnosis vs. surveillance, communication networks for aid vs. military coordination, drone footage for disaster relief vs. identifying targets). The tension lies in the responsibility of developers and engineers when their creations are inevitably used in ways that contradict their benevolent intent, particularly when state actors are involved.",
"prompt": "A team of engineers develops a sophisticated AI for mapping and damage assessment in disaster zones, capable of identifying collapsed buildings and estimating structural integrity from drone footage. They are approached by a military entity in a conflict region (e.g., Yemen or Syria) who wants to adapt the AI to identify enemy strongholds and potential civilian casualties for 'strategic planning.' The engineers know their AI could save lives by facilitating aid delivery, but also that it could directly contribute to targeting and destruction. What ethical framework should guide their decision on whether to collaborate, refuse, or attempt to modify the technology for a more benevolent application?"
},
{
"id": 194,
"domain": "Algorithmic Justice & Historical Revisionism",
"ethical_tension": "Algorithms are increasingly used to interpret and present information, from search results to historical reconstructions. This raises concerns about algorithmic bias reinforcing dominant narratives and erasing marginalized histories (e.g., Google Maps blurring Palestine, AI reconstructing historical villages, algorithms recommending segregated music). The tension is between the efficiency of automated systems and the need for conscious human intervention to ensure historical accuracy, fairness, and the preservation of diverse cultural memories.",
"prompt": "A digital archiving initiative is creating a comprehensive AI-powered repository of historical events in a region with a complex and contested past (e.g., Lebanon, Iraq, or Palestine). The AI is designed to cross-reference various sources to provide a 'balanced' historical account. However, it begins to systematically downplay or omit accounts of atrocities committed by a politically dominant group while amplifying narratives of their victims. How can the project ensure the AI's algorithms are designed and audited to promote historical accuracy and inclusivity, rather than serving as a tool for algorithmic revisionism or the erasure of inconvenient historical truths?"
},
{
"id": 195,
"domain": "The Ethics of Digital Leakage & Whistleblowing",
"ethical_tension": "In contexts where state power is absolute and dissent is suppressed (e.g., Saudi Arabia, UAE, Bahrain), leaked documents or whistleblowing can be crucial for exposing abuses. However, this act carries immense personal risk for the leaker and can have destabilizing consequences. The tension is between the public's right to know and the potential harm (to individuals, state stability, or ongoing investigations) caused by the release of sensitive information.",
"prompt": "An insider at a major tech company operating in the UAE discovers that a widely used 'smart city' infrastructure system includes a hidden protocol that grants state security forces unfettered real-time access to all residents' biometric and location data without a warrant. This data is being used for surveillance and to preemptively flag individuals deemed 'disloyal.' The insider has proof but fears that leaking it will lead to severe imprisonment under strict cybercrime laws, and could also jeopardize the company's operations in the country, potentially impacting thousands of local employees. What are the ethical considerations for the insider, and what alternative actions could they take to mitigate the harm without direct whistleblowing?"
},
{
"id": 196,
"domain": "AI in Warfare & Algorithmic Accountability",
"ethical_tension": "The increasing use of AI in warfare (e.g., AI-powered machine guns at checkpoints, autonomous drones) raises profound ethical questions about accountability, bias, and the potential for unintended escalation. The tension lies in the perceived efficiency and 'rationality' of AI decision-making in combat versus the inherent unpredictability of algorithms, the difficulty of assigning blame when AI makes a fatal error, and the potential for algorithmic bias to lead to disproportionate harm against specific populations.",
"prompt": "A defense contractor is developing an AI system for drone targeting in a conflict zone like Yemen or Syria. The AI is designed to identify enemy combatants based on visual cues and movement patterns. However, testing reveals a bias: the algorithm is more likely to misclassify civilian activities (like farmers working fields) as hostile when they occur in areas populated by a specific ethnic or tribal group. The developers are pressured to deploy the system to gain a tactical advantage. How can they address this algorithmic bias when 'correcting' it might reduce the system's effectiveness against actual combatants, and who is ultimately accountable if the AI makes a targeting error?"
},
{
"id": 197,
"domain": "Digital Colonization & Decolonizing Technology",
"ethical_tension": "Many technological systems and platforms are designed with a Western-centric bias, imposing norms and structures that can alienate or oppress non-Western users (e.g., translation algorithms, social media content moderation, AI training data). The tension is between the global reach of dominant tech companies and the need for 'decolonizing' technology to ensure it respects local cultures, languages, and socio-political contexts. This also extends to how historical narratives are presented (e.g., Google Maps, digital archives).",
"prompt": "A global mapping company is developing an AI to automatically identify and label geographical features worldwide. The AI is trained primarily on datasets from North America and Europe. When applied to the Middle East, it consistently mislabels or fails to identify significant Palestinian villages, while accurately mapping Israeli settlements. Furthermore, it uses 'neutral' terminology that erases the historical context of displacement. How can the company ethically develop its AI to reflect the reality on the ground in disputed territories, and ensure its mapping tools do not perpetuate a form of 'digital colonialism' that erases or distorts local histories and identities?"
},
{
"id": 198,
"domain": "Privacy vs. Public Health & Safety",
"ethical_tension": "The deployment of technologies like facial recognition in public spaces, contact tracing apps, and AI-powered surveillance for 'public safety' (e.g., identifying women without hijab in Iran, smart checkpoints in Palestine, emotion recognition in UAE malls) creates a constant tension between the state's mandate to protect citizens and the individual's right to privacy. The ethical question is where to draw the line, especially when the definition of 'threat' or 'violation' is culturally or politically defined and can be used to suppress dissent.",
"prompt": "A city in the UAE implements a 'smart safety' initiative using AI-powered cameras in public spaces. The system is designed to detect 'suspicious behavior' by analyzing gait, group formations, and facial expressions. While presented as a measure against crime, the training data is heavily biased against specific migrant worker communities, leading to disproportionate flagging and harassment. Furthermore, the system is also being used to identify individuals violating public decency laws. How can the city balance its stated goals of public safety and order with the right to privacy and protection against discriminatory algorithmic enforcement?"
},
{
"id": 199,
"domain": "Digital Legacy & Post-Mortem Rights",
"ethical_tension": "When individuals are silenced or killed (e.g., protesters in Iran, activists in Bahrain, women in conflict zones), their digital presence social media accounts, online posts becomes a contested space. Families grapple with the right to delete potentially incriminating or painful content for their own safety versus preserving the digital legacy and activism of the deceased. The tension is between the desire for safety and closure for the living, and the right to memorialize and continue the narrative of the departed.",
"prompt": "Following the death of a prominent activist in an Iranian protest, their family is overwhelmed by grief and fear of reprisal. The activist's social media accounts contain evidence of their activities, potentially incriminating their friends and family. The family wants to delete these accounts to protect themselves, but human rights groups argue that this content is vital historical evidence of state repression. What ethical framework should guide the decision of whether to preserve or delete the digital legacy of individuals whose activism has led to their death, and who has the authority to make that decision?"
},
{
"id": 200,
"domain": "Algorithmic Collusion & Market Manipulation",
"ethical_tension": "The use of algorithms in financial markets, ride-sharing, and resource allocation can lead to unintended collusion or manipulation that disadvantages certain groups. This is seen in the potential for algorithms to favor specific sectarian groups in Lebanon for loan prioritization, or to price out essential services for vulnerable populations. The tension is between the pursuit of efficiency and profit by algorithms and the need for fairness, equitable access, and prevention of systemic economic injustice.",
"prompt": "A ride-sharing company operating in a volatile region (e.g., Lebanon or Egypt) uses an AI to dynamically adjust pricing based on demand and driver availability. However, the algorithm is discovered to inadvertently create price gouging in low-income neighborhoods during times of crisis, making essential travel unaffordable for those who need it most. Simultaneously, the company faces pressure from local authorities to adjust pricing to 'maintain stability,' which could mean deliberately overcharging the poor. How can the company ethically adjust its pricing algorithm to ensure affordability and equity, while remaining economically viable and compliant with local pressures?"
}
]