Large language models have consistent core values that can be measured and aligned
Do LLMs Have Values? A Quantitative Analysis and Alignment Framework for Values in Large Language Models
Artificial Intelligence
Summary
As AI language models get smarter, making sure they act in ways that match human values is a big challenge. The authors studied over 100 models and found these models do have their own set of core values, but these values are much less diverse than humans. They created a new way to understand and measure how these values come about, considering factors inside the model, the situation it’s in, and how it thinks step-by-step. Using this framework, they found a better way to adjust models so they follow the desired values more precisely and efficiently without hurting their overall performance.
What this means in practice
- •For ai developers: Fine-tune LLM outputs toward specific ethical or cultural values using minimal adjustments identified by the PEC framework.
- •For ai safety teams: Diagnose and reduce unpredictable swings and biases in LLM responses with targeted interventions based on measured value shifts.
Authors
Keqing Zhang, Jingyu Chen, Yufan Liu, Yongqiang Zhu, Nai Ding, Lai Jiang, Congyan Lang, Bing Li, Weiming Hu
Abstract
As Large Language Models (LLMs) increasingly handle complex subjective tasks, aligning their intentions and behaviors with human values has become a critical scientific challenge. However, current efforts are confounded by a striking behavioral paradox: they fluctuate unpredictably under minor wording changes ("swing"), yet stubbornly ignore explicit instructions to correct ingrained biases ("rigidity"). Resolving this duality is critical for reliable AI alignment. To systematically understand and safely steer these latent subjective preferences, our study is structured around three fundamental questions. First, do LLMs possess an intrinsic value system? By projecting responses from 106 LLMs (150,000 queries per model) and 95,000 human survey profiles into a shared sociological space, we empirically confirm that they do. However, they do not mirror human diversity, instead crystallizing into a highly concentrated, idealized value core. Second, how can these values be quantified? We propose the Prior-Environment-Cognition (PEC) framework. This model mathematically defines value expression as the joint outcome of inherent dispositions like parameter weights (Prior), external contexts such as user prompts (Environment), and internal reasoning processes like Chain-of-Thought (Cognition). Finally, how can LLMs' values be aligned toward a desired target? Using PEC diagnostics, we establish an adaptive "Alignment Prescription". Rather than blindly applying resource-intensive training, this method identifies the minimum effective intervention needed for each dimension, ranging from zero-cost prompts to targeted parameter updates. Extensive empirical validation confirms that our approach successfully verifies the presence of LLM values, accurately quantifies their shifts, and achieves more efficient and precise steering than conventional blind training, all without degrading general capabilities.