Munich Center for Machine Learning published a new preprint analyzing 89,253 outputs from 12 LLMs in English, French, and Chinese across 18 occupations and three task conditions.
Public source
Publisher name
Public post
🌍 How culturally grounded are multilingual LLMs—really? ➡️ New preprint: 𝐁𝐞𝐲𝐨𝐧𝐝 𝐒𝐮𝐫𝐟𝐚𝐜𝐞 𝐂𝐮𝐞𝐬: 𝐃𝐢𝐬𝐞𝐧𝐭𝐚𝐧𝐠𝐥𝐢𝐧𝐠 𝐒𝐨𝐜𝐢𝐨𝐜𝐮𝐥𝐭𝐮𝐫𝐚𝐥 𝐒�…
Company
Munich Center for Machine Learning
Making AI potential accessible for science, industry and society
- Industry
- Research Services
- Location
- München, DE
- Company size
- 51–200 employees
About Munich Center for Machine Learning
The Munich Center for Machine Learning is a joint research initiative of Ludwig-Maximilians-Universität München (LMU) and Technische Universität München (TUM). It is one of six nationwide AI centers in Germany to receive permanent funding as part of the German and Bavarian government's AI strategy. Our vision is to strengthen regional, national and international competence in AI and to make the corresponding potential accessible to users from science and industry.
See moreLatest activity
Latest activity from Munich Center for Machine Learning
5 signals
Research & Knowledge
Munich Center for Machine Learning conducted joint work with Valentina Bosetti and Laura Diaz Anadon on applying causal machine learning to assess and improve cleantech policy design.
Research & Knowledge
Munich Center for Machine Learning updated its preprint comparing human-alone vs LLM+human diagnostic assistance and analyzed 42 clinical studies covering 101 effect sizes.
Research & Knowledge
Munich Center for Machine Learning PI Julia A. Schnabel develops AI models to detect medical imaging anomalies such as cancer metastases and cardiovascular problems.
Discover more
Similar signals
Similar public activity from other companies.
Research & Knowledge
NEC Laboratories America, Inc.
NEC Laboratories America, Inc. published a paper titled Mind the Gap in Cultural Alignment: Task-Aware Culture Management for Large Language Models on July 7, 2026.
Research & Knowledge
alphaXiv
alphaXiv published a paper identifying that LLMs fail during pretraining due to disjoint token spaces, even between two identical copies of English, suggesting that mapping languages into a shared semantic token space improves cross-lingual learning by up to 14x.
Research & Knowledge
Fraunhofer FIT Generative AI Lab
Fraunhofer FIT Generative AI Lab published a recent paper introducing the concept of robotoid humanness in human-AI interaction.
Research & Knowledge
Twisters Management Consulting LLC
Twisters Management Consulting LLC published a paper investigating machine psychology and persona-based dialogue generation, demonstrating that large language models exhibit a systematic bias toward persona overuse.
Research & Knowledge
Google DeepMind