Fanzhi Zeng

Jiaming Ji, Tianyi Qiu, Boyuan Chen et al.

AI alignment aims to make AI systems behave in line with human intentions and values. As AI systems grow more capable, so do risks from misalignment. To provide a comprehensive and up-to-date overview of the alignment field, in this survey, we delve into the core concepts, methodology, and practice of alignment. First, we identify four principles as the key objectives of AI alignment: Robustness, Interpretability, Controllability, and Ethicality (RICE). Guided by these four principles, we outline the landscape of current alignment research and decompose them into two key components: forward alignment and backward alignment. The former aims to make AI systems aligned via alignment training, while the latter aims to gain evidence about the systems' alignment and govern them appropriately to avoid exacerbating misalignment risks. On forward alignment, we discuss techniques for learning from feedback and learning under distribution shift. On backward alignment, we discuss assurance techniques and governance practices. We also release and continually update the website (www.alignmentsurvey.com) which features tutorials, collections of papers, blog posts, and other resources.

Hannah Rose Kirk, Liu Leqi, Fanzhi Zeng et al.

Personalisation is a standard feature of conversational AI systems used by millions; yet, the efficacy of personalisation methods is often evaluated in academic research using simulated users rather than real people. This raises questions about how users and their simulated counterparts differ in interaction patterns and judgements, as well as whether personalisation is best achieved through context-based prompting or weight-based fine-tuning. Here, in a large-scale within-subject experiment, we re-recruit 530 participants from 52 countries two years after they gave their preferences in the PRISM dataset (Kirk et al., 2024) to evaluate personalised and non-personalised language models in blinded multi-turn conversations. We find preference fine-tuning (P-DPO, Li et al., 2024) significantly outperforms both a generic model and personalised prompting but adapting to individual preference data yields marginal gains over training on pooled preferences from a diverse population. Beyond length biases, fine-tuning amplifies sycophancy and relationship-seeking behaviours that people reward in short-term evaluations but which may introduce deleterious long-term consequences. Replicating this within-subject experiment with simulated users recovers aggregate model hierarchies but simulators perform far below human self-consistency baselines for individual judgements, discuss different topics, exhibit amplified position biases, and produce feedback dynamics that diverge from humans.

Jincheng Cao, Fanzhi Zeng, Leqi Liu et al.

Knowledge distillation is a key technique for transferring the capabilities of large language models (LLMs) into smaller, more efficient student models. Existing distillation approaches often overlook two critical factors: the learning order of training data and the capacity mismatch between teacher and student models. This oversight limits distillation performance, as manifested by the counter-intuitive phenomenon where stronger teachers fail to produce better students. In this work, we propose Curriculum Learning-Guided Progressive Distillation (CLPD), a unified framework that explicitly accounts for both factors by aligning data difficulty with teacher strength. CLPD constructs an explicit curriculum by organizing training examples from easy to hard, while simultaneously applying an implicit curriculum over supervision signals by progressively scheduling teachers of increasing capacity. Our framework is modular and can be integrated into standard distillation algorithms with minimal overhead. Empirical results on the reasoning benchmarks demonstrate that CLPD consistently outperforms standard distillation, data ordering alone, and teacher scheduling alone across multiple settings. These findings highlight the importance of jointly considering data ordering and teacher capacity when distilling reasoning abilities into small language models.

Tianyi Qiu, Fanzhi Zeng, Jiaming Ji et al.

Existing alignment methods share a common topology of information flow, where reward information is collected from humans, modeled with preference learning, and used to tune language models. However, this shared topology has not been systematically characterized, nor have its alternatives been thoroughly explored, leaving the problems of low data efficiency and unreliable generalization unaddressed. As a solution, we introduce a theory of reward generalization in reinforcement learning from human feedback (RLHF), focusing on the topology of information flow at both macro and micro levels. At the macro level, we portray the RLHF information flow as an autoencoding process over behavior distributions, formalizing the RLHF objective of distributional consistency between human preference and model behavior. At the micro level, we present induced Bayesian networks to model the impact of dataset topologies on reward generalization. Combining analysis on both levels, we propose reward modeling from tree-structured preference information. It is shown to reduce reward uncertainty by up to $Θ(\log n/\log\log n)$ times compared to baselines, where $n$ is the dataset size. Validation on three NLP tasks shows that it achieves an average win rate of 65% against baselines, thus improving reward generalization for free via topology design, while reducing the amount of data requiring annotation.

Fanzhi Zeng, Siqi Wang, Chuzhao Zhu et al.

How to construct an interpretable autonomous driving decision-making system has become a focal point in academic research. In this study, we propose a novel approach that leverages large language models (LLMs) to generate executable, rule-based decision systems to address this challenge. Specifically, harnessing the strong reasoning and programming capabilities of LLMs, we introduce the ADRD(LLM-Driven Autonomous Driving Based on Rule-based Decision Systems) framework, which integrates three core modules: the Information Module, the Agents Module, and the Testing Module. The framework operates by first aggregating contextual driving scenario information through the Information Module, then utilizing the Agents Module to generate rule-based driving tactics. These tactics are iteratively refined through continuous interaction with the Testing Module. Extensive experimental evaluations demonstrate that ADRD exhibits superior performance in autonomous driving decision tasks. Compared to traditional reinforcement learning approaches and the most advanced LLM-based methods, ADRD shows significant advantages in terms of interpretability, response speed, and driving performance. These results highlight the framework's ability to achieve comprehensive and accurate understanding of complex driving scenarios, and underscore the promising future of transparent, rule-based decision systems that are easily modifiable and broadly applicable. To the best of our knowledge, this is the first work that integrates large language models with rule-based systems for autonomous driving decision-making, and our findings validate its potential for real-world deployment.

Femi Bello, Anubrata Das, Fanzhi Zeng et al.

It has been hypothesized that neural networks with similar architectures trained on similar data learn shared representations relevant to the learning task. We build on this idea by extending the conceptual framework where representations learned across models trained on the same data can be expressed as linear combinations of a \emph{universal} set of basis features. These basis features underlie the learning task itself and remain consistent across models, regardless of scale. From this framework, we propose the \textbf{Linear Representation Transferability (LRT)} Hypothesis -- that there exists an affine transformation between the representation spaces of different models. To test this hypothesis, we learn affine mappings between the hidden states of models of different sizes and evaluate whether steering vectors -- directions in hidden state space associated with specific model behaviors -- retain their semantic effect when transferred from small to large language models using the learned mappings. We find strong empirical evidence that such affine mappings can preserve steering behaviors. These findings suggest that representations learned by small models can be used to guide the behavior of large models, and that the LRT hypothesis may be a promising direction on understanding representation alignment across model scales.