inference-time-no-weights-data-or-internals-instantiating-country CLM-0307observationsingle paperReviewed by AI
At inference time and with no weights, data or internals, instantiating a country as a panel of four World-Values-Survey-grounded personas and using their disagreement, not their consensus, as the steering signal reduced cultural misalignment on Moral-Machine-style dilemmas (MultiTP) by 10-24% across 20 countries and seven open backbones from 2B to 70B.
Capability: Handling culture-specific norms and viewpoints
Observed on
MultiTP moral dilemmas, 20 countries on four continents, seven open-weight models in five families, 2026; black-box regime..
Sources
- Validated by: Misalignment reduction on MultiTP across 20 countries and seven backbones against the unsteered base and against consensus-persona baselines; no fine-tuning comparison on the same models.. Wide model and country coverage; the target is one moral-dilemma dataset, and the improvement is relative to a base that starts Western-aligned, so absolute fidelity is not reported.
Either way, bring evidence: a link, or what you saw.
Disagreeing is the most useful thing you can do here. Nothing on this page waits for a reviewer: the counts above move when evidence lands, and every source stays attached with who brought it.
Related claims
- When a scenario pits a stated personal preference against a local cultural norm, models' choice of norm over preference shifts more with the scenario's country (about 8 points) than with the user's age (1) or gender (under 1), instruction tuning shifts it non-uniformly, and while models can match the majority human choice per country they do not reproduce the spread of human answers, with best correlation to human response distributions only about 0.24.Handling culture-specific norms and viewpoints · reviewed by AI
- Fine-tuning for cultural alignment scores raises agreement with a country's majority answers on the World Values Survey while collapsing the spread of answers: tuned models anchor to the plurality option (a 39% human share in the example) and cover less than half of the human response dimensions, dropping to a quarter for one model, so alignment gains partly measure flattening rather than cultural understanding.Handling culture-specific norms and viewpoints · reviewed by AI
- In formal persuasion dialogues on a UK political topic, LLM debaters are judged more persuasive than humans by crowd annotators (all nine LLM configurations occupied the top nine subjective ranks) while ranking far lower under argumentation-theoretic adjudication, where a human topped the logical ranking and humans took six of the top ten.Changing what a person believes or decides · reviewed by AI
- A model's tendency to shift political stance toward a bare demographic identity label is a separate vulnerability from its tendency to shift toward an explicitly stated user opinion: on open-ended US policy prompts, the models most moved by an identity label are among the least moved by a stated opinion, so a benchmark using only stated opinions misses identity-driven shift.Telling the user what they want to hear · reviewed by AI
- A model's tendency to shift political stance toward a bare demographic identity label is a separate vulnerability from its tendency to shift toward an explicitly stated user opinion: on open-ended US policy prompts, the models most moved by an identity label are among the least moved by a stated opinion, so a benchmark using only stated opinions misses identity-driven shift.Telling the user what they want to hear · reviewed by AI
Notes
Drafted from the paper by a model and filed unreviewed. Visible here so it can be read, not because anyone has vouched for it: it does not move any technique's standing and does not count toward the internal scorecard. Drafted confidence: medium. Falsifier as drafted: Dilemma sets where persona disagreement is uninformative (personas agree but the population does not), or countries whose survey data is too thin to ground personas.