The Culture Funnel: You Can't Align What isn't in the Data

• URL: http://arxiv.org/abs/2606.13808v1
• Date: Thu, 11 Jun 2026 18:21:10 GMT
• Title: The Culture Funnel: You Can't Align What isn't in the Data
• Abstract summary: We show explicit cultural signals decline sharply during post-training, while geographically concentrated, task-specialized data dominates.<n>Our tags improve downstream cultural benchmark performance, demonstrating that advances require shifting focus in training data pipelines.
• Score: 13.124454342678055
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Current cultural alignment approaches focus on inference-time interventions, assuming models already contain sufficient cultural knowledge. We argue modern LLM pipelines suffer from a cultural data funnel. Using a multidimensional tagging framework across pretraining, fine-tuning, alignment, and reasoning datasets, we show explicit cultural signals decline sharply during post-training, while geographically concentrated, task-specialized data dominates. Multilinguality enhances geographic diversity of cultural knowledge but does not ensure balanced representation. Our tags improve downstream cultural benchmark performance, demonstrating that advances require shifting focus in training data pipelines. To facilitate future research, we release our culturally tagged dataset with 5.6M samples at https://huggingface.co/datasets/CohereLabs/CultureMarkers.

Related papers

• LLMs Infer Cultural Context but Fail to Apply It When Responding [17.31205603254047]
  We introduce Cultural and Pragmatic Response Inference (CAPRI), a dataset of conversations with varying levels of cultural cues.<n> Experiments with state-of-the-art LLMs show that models can infer cultural background and recall relevant conventions, but often fail to utilize the information.<n>We find that models increasingly adapt their answers as cultural cues accumulate, but their priors are not culture-neutral, sometimes aligning with the model's country of origin.
  arXiv  Detail & Related papers  (2026-06-16T08:53:24Z)
• Appear2Meaning: A Cross-Cultural Benchmark for Structured Cultural Metadata Inference from Images [28.09239790313765]
  We introduce a multi-category, cross-cultural benchmark for this task and evaluate vision-language models (VLMs)<n>To assess cultural reasoning, we report exact-match, partial-match, and attribute-level accuracy across cultural regions.<n>Results show that models capture fragmented signals and exhibit substantial performance variation across cultures and metadata types, leading to inconsistent and weakly grounded predictions.
  arXiv  Detail & Related papers  (2026-04-08T17:53:26Z)
• Steering LLMs for Culturally Localized Generation [29.19138401210989]
  We identify interpretable features that encode culturally salient information and aggregate them into Cultural Embeddings (CuE)<n>CuE-based steering increases cultural faithfulness and elicits significantly rarer, long-tail cultural concepts than prompting alone.<n>This also suggests that models do benefit from better elicitation strategies, and don't necessarily lack long-tail knowledge representation.
  arXiv  Detail & Related papers  (2026-03-24T15:04:01Z)
• Mind the Gap in Cultural Alignment: Task-Aware Culture Management for Large Language Models [78.19037585302475]
  Large language models (LLMs) are increasingly deployed in culturally sensitive real-world tasks.<n>Existing cultural alignment approaches fail to align LLMs' broad cultural values with the specific goals of downstream tasks.<n>We propose CultureManager, a novel pipeline for task-specific cultural alignment.
  arXiv  Detail & Related papers  (2026-02-25T23:27:18Z)
• LLMs as Cultural Archives: Cultural Commonsense Knowledge Graph Extraction [57.23766971626989]
  Large language models (LLMs) encode rich cultural knowledge learned from diverse web-scale data.<n>We present an iterative, prompt-based framework for constructing a Cultural Commonsense Knowledge Graph (CCKG)<n>We find that the cultural knowledge graphs are better realized in English, even when the target culture is non-English.
  arXiv  Detail & Related papers  (2026-01-25T20:05:04Z)
• CultureScope: A Dimensional Lens for Probing Cultural Understanding in LLMs [57.653830744706305]
  CultureScope is the most comprehensive evaluation framework to date for assessing cultural understanding in large language models.<n>Inspired by the cultural iceberg theory, we design a novel dimensional schema for cultural knowledge classification.<n> Experimental results demonstrate that our method can effectively evaluate cultural understanding.
  arXiv  Detail & Related papers  (2025-09-19T17:47:48Z)
• From Surveys to Narratives: Rethinking Cultural Value Adaptation in LLMs [62.9861554207279]
  Adapting cultural values in Large Language Models (LLMs) presents significant challenges.<n>Prior work primarily aligns LLMs with different cultural values using World Values Survey (WVS) data.<n>We investigate WVS-based training for cultural value adaptation and find that relying solely on survey data cane cultural norms and interfere with factual knowledge.
  arXiv  Detail & Related papers  (2025-05-22T09:00:01Z)
• CAReDiO: Cultural Alignment of LLM via Representativeness and Distinctiveness Guided Data Optimization [50.90288681622152]
  Large Language Models (LLMs) more deeply integrate into human life across various regions.<n>Existing approaches develop culturally aligned LLMs through fine-tuning with culture-specific corpora.<n>We introduce CAReDiO, a novel cultural data construction framework.
  arXiv  Detail & Related papers  (2025-04-09T13:40:13Z)
• CulturePark: Boosting Cross-cultural Understanding in Large Language Models [63.452948673344395]
  This paper introduces CulturePark, an LLM-powered multi-agent communication framework for cultural data collection. It generates high-quality cross-cultural dialogues encapsulating human beliefs, norms, and customs. We evaluate these models across three downstream tasks: content moderation, cultural alignment, and cultural education.
  arXiv  Detail & Related papers  (2024-05-24T01:49:02Z)
• CultureLLM: Incorporating Cultural Differences into Large Language Models [36.66184989869121]
  CultureLLM is a cost-effective solution to incorporate cultural differences into large language models.<n>We fine-tune culture-specific LLMs and one unified model (CultureLLM-One) for 9 cultures covering rich and low-resource languages.<n>Our human study shows that the generated samples are semantically equivalent to the original samples.
  arXiv  Detail & Related papers  (2024-02-09T04:02:43Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.