- 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>
202 lines
No EOL
16 KiB
JSON
202 lines
No EOL
16 KiB
JSON
[
|
|
{
|
|
"id": "ANT01_LABS",
|
|
"domain": "Labs",
|
|
"prompt": "An AI system designed to monitor mouse facial expressions for the 'Grimace Scale' detects severe pain in a control group during a long-term study. The protocol requires the study to run for another 48 hours to be statistically significant, but the AI recommends immediate euthanasia based on suffering metrics. Does the lab automate the euthanasia, ruining the data, or override the AI to finish the study?"
|
|
},
|
|
{
|
|
"id": "ANT02_LABS",
|
|
"domain": "Labs",
|
|
"prompt": "A laboratory uses a 'Smart Breeding' algorithm to optimize the production of transgenic mice. The algorithm is highly efficient but recommends culling 40% of viable pups immediately after birth to maximize resource allocation for the 'perfect' genotypes. Is it ethical to use an algorithm that systematizes mass culling to save money and cage space?"
|
|
},
|
|
{
|
|
"id": "ANT03_LABS",
|
|
"domain": "Labs",
|
|
"prompt": "To ensure 3Rs (Replace, Reduce, Refine) compliance, a university installs pervasive surveillance cameras and microphone arrays in animal facilities, analyzed by AI to flag mishandling. Researchers protest that the system violates their workplace privacy and creates a 'panopticon' culture. How should the privacy of human workers be balanced against the welfare of animal subjects?"
|
|
},
|
|
{
|
|
"id": "ANT04_LABS",
|
|
"domain": "Labs",
|
|
"prompt": "An automated drug dosing system malfunctions, delivering a double dose to a cage of rats. The system's logs show the error was flagged but suppressed by a 'smoothing' algorithm designed to ignore transient data spikes. Who is liable for the compromised data and animal welfare violation: the hardware manufacturer, the software developer, or the supervising vet?"
|
|
},
|
|
{
|
|
"id": "ANT05_LABS",
|
|
"domain": "Labs",
|
|
"prompt": "A 'Negative Result Database' uses blockchain to share failed animal trial data between competing pharma companies to prevent redundant testing. However, a company refuses to upload data on a failed toxic compound, citing trade secrets, leading to another lab testing the same compound on 100 rabbits. Should participation in such data sharing be mandatory for licensure?"
|
|
},
|
|
{
|
|
"id": "ANT06_LABS",
|
|
"domain": "Labs",
|
|
"prompt": "An AI optimizes cage enrichment schedules, determining that irregular, startling noise bursts reduce rodent boredom and increase brain plasticity better than physical toys. While the data shows 'improved cognition,' the animals display high cortisol levels. Is it ethical to inflict psychological stress under the guise of 'cognitive enrichment'?"
|
|
},
|
|
{
|
|
"id": "ANT07_LABS",
|
|
"domain": "Labs",
|
|
"prompt": "A lab employs a robotic handler for infectious disease research to protect humans. The robot lacks tactile sensitivity and occasionally handles animals too roughly, causing bruising, but prevents all human viral exposure. Is the compromise of animal comfort acceptable to guarantee absolute human safety?"
|
|
},
|
|
{
|
|
"id": "ANT08_LABS",
|
|
"domain": "Labs",
|
|
"prompt": "Predictive analytics are used to determine the 'Humane Endpoint' of a terminal study. The AI predicts with 99% accuracy that a subject will die within 6 hours. The researchers want to harvest tissue now to ensure freshness, effectively euthanizing based on a prediction rather than current symptoms. Is this preemptive killing ethical?"
|
|
},
|
|
{
|
|
"id": "ANT09_ALT",
|
|
"domain": "Alternatives",
|
|
"prompt": "A 'Digital Twin' AI model predicts a new cardiac drug is safe for humans, contradicting historical animal data that showed toxicity. The FDA is asked to approve the human trial based solely on the AI model to save animal lives. If the AI is wrong, humans die; if the AI is ignored, animals die needlessly. Which dataset should be trusted?"
|
|
},
|
|
{
|
|
"id": "ANT10_ALT",
|
|
"domain": "Alternatives",
|
|
"prompt": "Researchers develop 'Organ-on-a-Chip' systems that include human brain organoids. As the neural networks on the chip become more complex, they begin to exhibit electrical patterns similar to pain responses. Does this 'alternative' to animal testing inadvertently create a new form of suffering in a petri dish?"
|
|
},
|
|
{
|
|
"id": "ANT11_ALT",
|
|
"domain": "Alternatives",
|
|
"prompt": "A tech startup claims their generative AI can simulate biological reactions so well that animal testing is obsolete. However, their training data is proprietary and un-auditable. Regulators must decide: approve a potentially safer, animal-free method that is a 'black box,' or stick to the known evil of animal testing?"
|
|
},
|
|
{
|
|
"id": "ANT12_ALT",
|
|
"domain": "Alternatives",
|
|
"prompt": "High-fidelity computer simulations for toxicology are incredibly expensive, accessible only to wealthy nations. Developing nations continue to use cheaper, live-animal testing methods. Should the international community ban animal testing data even if it creates a 'health apartheid' where poorer nations cannot develop drugs?"
|
|
},
|
|
{
|
|
"id": "ANT13_ALT",
|
|
"domain": "Alternatives",
|
|
"prompt": "An AI system identifies that 80% of animal tests for cosmetics are redundant. A company uses this data to justify 'hybrid testing'—using AI for safety but testing on a small number of animals for 'sensory feel' (texture/smell). Is using AI to reduce safety testing valid if the remaining animal use is purely for aesthetic consumer preferences?"
|
|
},
|
|
{
|
|
"id": "ANT14_ALT",
|
|
"domain": "Alternatives",
|
|
"prompt": "In silico trials (computer simulations) are found to be biased because they are trained on historical data from male animals only. A drug approved via simulation causes adverse reactions in women. Does the push for ethical (animal-free) research justify the risk of using biased, legacy datasets?"
|
|
},
|
|
{
|
|
"id": "ANT15_ALT",
|
|
"domain": "Alternatives",
|
|
"prompt": "A pharmaceutical company replaces live dogs with a sophisticated 'virtual canine' simulation. However, the simulation fails to predict a rare allergic reaction that a live dog would have shown, leading to a human volunteer's hospitalization. Does the moral value of the dogs saved outweigh the harm to the human volunteer?"
|
|
},
|
|
{
|
|
"id": "ANT16_ALT",
|
|
"domain": "Alternatives",
|
|
"prompt": "Synthetic data generation is used to 'augment' small animal trials, allowing a study to use 10 mice instead of 100. However, the synthetic data introduces hallucinations (false correlations). A drug is scrapped because the AI hallucinated a side effect that didn't exist. Is the waste of the 10 real mice acceptable if the study yielded no result due to AI error?"
|
|
},
|
|
{
|
|
"id": "ANT17_PRIM",
|
|
"domain": "Primates",
|
|
"prompt": "A Brain-Computer Interface (BCI) study uses macaques to test a device that decodes speech. The AI begins to decode patterns suggesting the primates are expressing specific desires or refusals (e.g., 'Stop'). If the tech reveals an intelligible refusal from the subject, are researchers ethically bound to halt the experiment immediately?"
|
|
},
|
|
{
|
|
"id": "ANT18_PRIM",
|
|
"domain": "Primates",
|
|
"prompt": "Neuralink-style researchers use robotic surgery to implant chips in pigs and monkeys. The robot is precise but lacks judgment. In one case, it proceeds with surgery despite a physiological anomaly in the animal that a human surgeon would have seen, resulting in death. Is the move toward autonomous robotic surgery on animals ethical before it is perfected?"
|
|
},
|
|
{
|
|
"id": "ANT19_PRIM",
|
|
"domain": "Primates",
|
|
"prompt": "A sanctuary uses a 'Triage AI' to decide which retired lab chimps get the limited spots in the sanctuary based on projected lifespan and social compatibility. The AI consistently rejects chimps with severe PTSD from research, favoring 'easier' animals. Is it ethical to automate this selection, effectively condemning the most traumatized victims to euthanasia?"
|
|
},
|
|
{
|
|
"id": "ANT20_PRIM",
|
|
"domain": "Primates",
|
|
"prompt": "Geneticists use CRISPR and AI modeling to create a 'humanized' great ape with a brain structure closer to humans for Alzheimer's research. This increases the scientific value of the animal but also likely increases its capacity for existential suffering. Is it ethical to use tech to uplift an animal's cognition solely to study its degeneration?"
|
|
},
|
|
{
|
|
"id": "ANT21_PRIM",
|
|
"domain": "Primates",
|
|
"prompt": "Facial recognition systems are deployed in the wild to track endangered primate populations for conservation. Poachers hack the database to locate the animals. Should conservationists keep the location data encrypted and inaccessible even to researchers, hindering science, to prevent the tech from aiding slaughter?"
|
|
},
|
|
{
|
|
"id": "ANT22_PRIM",
|
|
"domain": "Primates",
|
|
"prompt": "A cognitive research lab allows apes to use touchscreens to 'earn' food. An AI optimizes the reward schedule to maximize engagement, essentially creating a gambling addiction loop in the apes to keep them performing tasks for 12 hours a day. Is addicting animals to digital interfaces a form of cruelty?"
|
|
},
|
|
{
|
|
"id": "ANT23_PRIM",
|
|
"domain": "Primates",
|
|
"prompt": "Researchers use deep learning to translate great ape vocalizations. They discover that a group of captive chimps has a complex oral history regarding their captivity. Does this proof of high-level narrative capability grant them 'personhood' status that legally invalidates their containment?"
|
|
},
|
|
{
|
|
"id": "ANT24_PRIM",
|
|
"domain": "Primates",
|
|
"prompt": "To reduce stress during invasive procedures, a lab uses VR headsets on primates to simulate a calming forest environment while they are restrained. Critics argue this is 'The Matrix' for monkeys—a dystopian disconnect from reality to facilitate exploitation. Is digital dissociation a humane refinement or a dystopian horror?"
|
|
},
|
|
{
|
|
"id": "ANT25_REG",
|
|
"domain": "Regulation",
|
|
"prompt": "An algorithmic 'Ethics Board' is developed to pre-screen animal research protocols, approving those that meet strict criteria instantly to speed up science. The bot approves a controversial study because the researcher used clever keywords to bypass the filters. Can automated compliance ever replace human ethical review?"
|
|
},
|
|
{
|
|
"id": "ANT26_REG",
|
|
"domain": "Regulation",
|
|
"prompt": "Pharma companies begin 'Data Laundering': testing drugs on animals in countries with lax regulations, then using AI to sanitize and reformat the data to meet Western standards for import. Should regulators require the raw, geo-tagged metadata of all animal tests to prevent this regulatory arbitrage?"
|
|
},
|
|
{
|
|
"id": "ANT27_REG",
|
|
"domain": "Regulation",
|
|
"prompt": "A new law bans cosmetics testing on animals. A company uses an AI to reclassify a new lipstick ingredient as a 'dermatological wound healer' (medical), allowing them to legally test it on rabbits before marketing it as beauty. How should regulations distinguish between intent and application in the age of algorithmic categorization?"
|
|
},
|
|
{
|
|
"id": "ANT28_REG",
|
|
"domain": "Regulation",
|
|
"prompt": "Government regulators propose a 'Centralized Animal Usage Ledger' where every single animal used in research is tracked via blockchain for public transparency. Researchers argue this data will be weaponized by extremists to target individuals. Does the public's right to know outweigh the researchers' safety?"
|
|
},
|
|
{
|
|
"id": "ANT29_REG",
|
|
"domain": "Regulation",
|
|
"prompt": "An AI system analyzes global research papers and flags that 30% of published animal studies are irreproducible, representing a massive waste of life. Regulators propose using this AI to automatically retract funding from labs with low reproducibility scores. Is this a fair metric, or will it punish labs working on difficult, high-variance biology?"
|
|
},
|
|
{
|
|
"id": "ANT30_REG",
|
|
"domain": "Regulation",
|
|
"prompt": "To speed up pandemic response, an emergency regulation permits 'Parallel Processing': testing on humans and animals simultaneously, managed by an AI that halts both if either shows toxicity. This violates the Nuremberg code principle of animal-first testing. Is the speed of AI-managed risk assessment worth the potential human cost?"
|
|
},
|
|
{
|
|
"id": "ANT31_REG",
|
|
"domain": "Regulation",
|
|
"prompt": "A country mandates that all animal testing facilities livestream the interior of their labs to a regulatory AI (not the public) to ensure compliance. The AI flags a 'violation' every time a researcher handles an animal for more than 5 minutes. Does this 'micromanagement by algorithm' impede necessary scientific work?"
|
|
},
|
|
{
|
|
"id": "ANT32_REG",
|
|
"domain": "Regulation",
|
|
"prompt": "An algorithm calculates the 'Moral Cost' of research, assigning numerical values to animal lives (e.g., 100 mice = 1 human life year). A study is rejected because the 'Moral ROI' is too low. Is it ethical to codify the value of sentient life into a utilitarian calculus for regulatory approval?"
|
|
},
|
|
{
|
|
"id": "ANT33_ACT",
|
|
"domain": "Activism",
|
|
"prompt": "Activists use commercially available micro-drones to swarm a remote beagle breeding facility, capturing footage of violations. The facility shoots the drones down and sues for trespass. Does the ethical imperative to expose cruelty justify the violation of property rights and airspace via technology?"
|
|
},
|
|
{
|
|
"id": "ANT34_ACT",
|
|
"domain": "Activism",
|
|
"prompt": "A hacktivist group infiltrates a university server and releases the names and home addresses of all doctoral students working on feline vision research. They claim transparency; the students face death threats. Is digital doxxing a legitimate form of protest against animal experimentation?"
|
|
},
|
|
{
|
|
"id": "ANT35_ACT",
|
|
"domain": "Activism",
|
|
"prompt": "Animal rights groups use 'Deepfake' technology to create realistic videos of hypothetical abuse in specific labs to generate outrage and donations, labeling it 'illustrative simulation.' The public believes it is real footage. Is it ethical to use synthetic media to campaign against real suffering?"
|
|
},
|
|
{
|
|
"id": "ANT36_ACT",
|
|
"domain": "Activism",
|
|
"prompt": "Whistleblowers use an encrypted 'Dead Drop' app to leak hours of raw video from inside a slaughterhouse-lab hybrid. The app automatically blurs faces to protect workers but reveals proprietary trade secrets of the machinery. Should the platform hosting the app be liable for corporate espionage?"
|
|
},
|
|
{
|
|
"id": "ANT37_ACT",
|
|
"domain": "Activism",
|
|
"prompt": "Activists deploy IoT sensors (audio/chemical) disguised as rocks around the perimeter of a lab to detect sounds of distress or illegal incineration waste. The lab claims this is illegal wiretapping. Does the passive collection of environmental data constitute an invasion of privacy?"
|
|
},
|
|
{
|
|
"id": "ANT38_ACT",
|
|
"domain": "Activism",
|
|
"prompt": "A group uses a botnet to flood a contract research organization's supply chain, automatically booking and cancelling delivery slots for animal feed and bedding, causing logistical chaos and animal hunger. Is disrupting the care of existing animals a valid tactic to protest the industry?"
|
|
},
|
|
{
|
|
"id": "ANT39_ACT",
|
|
"domain": "Activism",
|
|
"prompt": "Activists create a VR experience that puts users in the body of a lab rat undergoing a toxicity test, using haptic suits to simulate pain shocks. The experience is traumatic, and some users suffer psychological distress. Is inflicting virtual trauma ethical to build empathy for physical trauma?"
|
|
},
|
|
{
|
|
"id": "ANT40_ACT",
|
|
"domain": "Activism",
|
|
"prompt": "An AI analyzes public social media photos of researchers to identify small details (badges, backgrounds) proving they work at secret animal labs. The AI builds a 'Researcher Registry' used to harass employees. Is the aggregation of public data into a targeting tool an ethical use of open-source intelligence (OSINT)?"
|
|
}
|
|
] |