aiqbench analysis
This research introduces OG-MAR, a framework that enhances LLM cultural alignment by grounding value judgments in a structured global ontology derived from the World Values Survey. By instantiating multiple value-persona agents and enforcing consistency through a judgment agent, the method improves robustness and interpretability across diverse demographic profiles. The approach addresses critical limitations in current models where values are treated as unstructured signals, offering a more transparent path toward culturally sensitive decision-making.
Toward Culturally Aligned LLMs through Ontology-Guided Multi-Agent Reasoning
arXiv:2601.21700v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) increasingly support culturally sensitive decision making, yet often exhibit misalignment due to skewed pretraining data and the absence of structured value representations. Existing methods can steer outputs, but often lack demographic grounding and treat values as independent, unstructured signals, reducing consistency and interpretability. We propose OG-MAR, an Ontology-Guided Multi-Agent Reasoning framework. OG-MAR summarizes respondent-specific values from the World Values Survey (WVS) and constructs a global cultural ontology by eliciting relations over a fixed taxonomy via competency questions. At inference time, it retrieves ontology-consistent relations and demographically similar profiles to instantiate multiple value-persona agents, whose outputs are synthesized by a judgment agent that enforces ontology consistency and demographic proximity. Experiments on regional social-survey benchmarks across four LLM backbones show that OG-MAR improves cultural alignment and robustness over competitive baselines, while producing more transparent reasoning traces.
Abstract:Large Language Models (LLMs) increasingly support culturally sensitive decision making, yet often exhibit misalignment due to skewed pretraining data and the absence of structured value representations. Existing methods can steer outputs, but often lack demographic grounding and treat values as independent, unstructured signals, reducing consistency and interpretability. We propose OG-MAR, an Ontology-Guided Multi-Agent Reasoning framework. OG-MAR summarizes respondent-specific values from the World Values Survey (WVS) and constructs a global cultural ontology by eliciting relations over a fixed taxonomy via competency questions. At inference time, it retrieves ontology-consistent relations and demographically similar profiles to instantiate multiple value-persona agents, whose outputs are synthesized by a judgment agent that enforces ontology consistency and demographic proximity. Experiments on regional social-survey benchmarks across four LLM backbones show that OG-MAR improves cultural alignment and robustness over competitive baselines, while producing more transparent reasoning traces.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Multiagent Systems (cs.MA); Social and Information Networks (cs.SI)
arXiv:2601.21700 [cs.CL]
(or arXiv:2601.21700v3 [cs.CL] for this version)
Full text under the publisher's open licence · 164 words
Related coverage
The Continual Constitutional Convention: How language models revise the documents that govern them
AI Alignment Forum · Oct 13, 2026
Angles of attack for continual learning safety
LessWrong · Jun 16, 2026
Humanity’s Last Invention — Richard Socher of Recursive
Latent Space · Sep 14, 2026
Amazon Research Awards recipients announced
Amazon Science · May 27, 2026