Austria

The increasing use of LLMs as substitutes for humans in &quot;aligning&quot; LLMs has raised questions about their ability to replicate human judgments and preferences, especially in ambivalent scenarios where humans disagree. This study examines the biases and limitations of LLMs in three roles: answer generator, judge, and debater. These roles loosely correspond to previously described alignment frameworks: preference alignment (judge) and scalable oversight (debater), with the answer generator reflecting the typical setting with user interactions. We develop a ``no-consensus&#39;&#39; benchmark by curating examples that encompass a variety of a priori ambivalent scenarios, each presenting two possible stances. Our results show that while LLMs can provide nuanced assessments when generating open-ended answers, they tend to take a stance on no-consensus topics when employed as judges or debaters. These findings underscore the necessity for more sophisticated methods for aligning LLMs without human oversight, highlighting that LLMs cannot fully capture human non-agreement even on topics where humans themselves are divided.

ACL 2025

Arbiters of Ambivalence: Challenges of using LLMs in No-Consensus tasks

alignment

fairness

## Welcome!
"I am excited to welcome you to this year’s edition of the Conference on Empirical Methods in Natural Language Processing! Importantly, it marks the 30th edition of EMNLP. With over 8,000 submissions, more than 3,000 accepted papers, and thousands of attendees, we have come a long way from that first
workshop, which had 14 accepted papers. As the field looks ahead, Suzhou is the fitting location for celebrating this milestone: rooted in a long literary tradition, yet modern and forward-looking, and home to a large share of the NLP community." 

*Message from the General Chair, Dirk Hovy*

EMNLP 2025

Celebrate 30 Years of EMNLP! 
EMNLP 2025 will be held in Suzhou, China from November 5th to November 9th, 2025.

The increasing use of LLMs as substitutes for humans in "aligning" LLMs has raised questions about their ability to replicate human judgments and preferences, especially in ambivalent scenarios where humans disagree. This study examines the biases and limitations of LLMs in three roles: answer generator, judge, and debater. These roles loosely correspond to previously described alignment frameworks: preference alignment (judge) and scalable oversight (debater), with the answer generator reflecting the typical setting with user interactions. We develop a ``no-consensus'' benchmark by curating examples that encompass a variety of a priori ambivalent scenarios, each presenting two possible stances. Our results show that while LLMs can provide nuanced assessments when generating open-ended answers, they tend to take a stance on no-consensus topics when employed as judges or debaters. These findings underscore the necessity for more sophisticated methods for aligning LLMs without human oversight, highlighting that LLMs cannot fully capture human non-agreement even on topics where humans themselves are divided.

poster

### Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

Message from the General Chair: 
*It is my great pleasure and honor to welcome you to the 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025), held in beautiful Vienna, Austria, from July 27 to August 1, 2025. ACL2025continues our field’s tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.* 
[Read more](https://drive.google.com/file/d/1GI_hvOpjswAuYdUTromfeDiPpCcqidwg/view?usp=sharing)

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to ACL 2025". Please be sure to check your spam email folder if you do not see an email confirmation right away.

Please log in

To access this event page, you need to log in with the **email address you registered with**. 

Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

It is challenging to update Large language models (LLMs) since real-world knowledge evolves. While existing Lifelong Knowledge Editing (LKE) methods efficiently update sequentially incoming edits, they often struggle to precisely overwrite the outdated knowledge with the latest one, resulting in conflicts that hinder LLMs from determining the correct answer. To address this $\textbf{S}$erial $\textbf{L}$ifelong $\textbf{K}$nowledge $\textbf{E}$diting (sLKE) problem, wepropose a novel Mixture-of-Knowledge-Experts scheme with an $\textbf{A}$ctivation-guided $\textbf{R}$outing $\textbf{M}$echanism (ARM), which assigns specialized experts to store domain-specific knowledge and ensures that each update completely overwrites old information with the latest data. Furthermore, we introduce a novel sLKE benchmark where answers to the same concept are updated repeatedly, to assess the ability of editing methods to refresh knowledge accurately. Experimental results on both LKE and sLKE benchmarks show that our ARM performs favorably against SOTA knowledge editing methods.

Serial Lifelong Editing via Mixture of Knowledge Experts

Integrating split learning with large language model fine-tuning (LLM-FT) enables secure collaboration between a trusted local client and a well-equipped remote server, but it is vulnerable to data reconstruction attacks (DRAs) that exploit transmitted activations and gradients. Current defense methods, like adding noise to activations or gradients, often sacrifice task-specific model performance under strict privacy constraints. This paper introduces DualGuard, a bidirectional defense mechanism against DRAs for split-based LLM-FT. DualGuard proposes a local warm-up parameter space transformation to alter client-side model parameters before training, using multi-task learning to strike a balance between privacy protection and model performance. Additionally, a global fine-tuning parameter space retention strategy prevents the model from reverting to vulnerable states during formal fine-tuning. Experiments show that DualGuard outperforms current defense methods against various DRAs, while maintaining task performance. Our code will be made publicly available.

DualGuard: A Parameter Space Transformation Approach for Bidirectional Defense in Split-Based LLM Fine-Tuning

Should LLMs generate language that makes them seem human? Human-like language might improve user experience, but might also lead to deception, overreliance, and stereotyping. Assessing these potential impacts requires a systematic way to measure human-like tone in LLM outputs. We introduce HumT and SocioT, metrics for human-like tone and other dimensions of social perceptions in text data based on relative probabilities from an LLM. By measuring HumT across preference and usage datasets, we find that users prefer less human-like outputs from LLMs in many contexts. HumT also offers insights into the perceptions and impacts of anthropomorphism: human-like LLM outputs are highly correlated with warmth, social closeness, femininity, and low status, which are closely linked to the aforementioned harms. We introduce DumT, a method using HumT to systematically control and reduce the degree of human-like tone while preserving model performance. DumT offers a practical approach for mitigating risks associated with anthropomorphic language generation.

HumT DumT: Measuring and controlling human-like language in LLMs

Large Language Models (LLMs) enhanced with external contexts, such as through retrieval-augmented generation (RAG), often face challenges in handling imperfect evidence. They tend to over-rely on external knowledge, making them vulnerable to misleading and unhelpful contexts. To address this, we propose the concept of context-robust LLMs, which can effectively balance internal knowledge with external context, similar to human cognitive processes. Specifically, context-robust LLMs should rely on external context only when lacking internal knowledge, identify contradictions between internal and external knowledge, and disregard unhelpful contexts. To achieve this goal, we introduce Grft, a lightweight and plug-and-play gated representation fine-tuning approach. Grft consists of two key components: a gating mechanism to detect and filter problematic inputs, and low-rank representation adapters to adjust hidden representations. By training a lightweight intervention function with only 0.0004% of model size on fewer than 200 examples, Grft can effectively adapt LLMs towards context-robust behaviors.

Towards Context-Robust LLMs: A Gated Representation Fine-tuning Approach

Large language models (LLMs) often struggle with zero-shot generalization, and several modular approaches have been proposed to address this challenge. Yet, we hypothesize that a key limitation remains: the entanglement of general knowledge and task-specific adaptations. To overcome this, we propose a modular framework that disentangles these components by constructing a library of task-specific LoRA modules alongside a general-domain LoRA. By subtracting this general knowledge component from each task-specific module, we obtain residual modules that focus more exclusively on task-relevant information. We call this approach general knowledge subtraction or GenKnowSub. Leveraging the refined task-specific modules and the Arrow routing algorithm, we dynamically select and combine modules for new inputs without additional training. Our studies on the Phi-3 model and standard Arrow as baselines reveal that using general knowledge LoRAs derived from diverse languages, including English, French, and German, yields consistent performance gains in both monolingual and cross-lingual settings across a wide set of benchmarks. Further experiments on Phi-2 reveal how GenKnowSub generalizes to a weaker LLM.

GenKnowSub: Improving Modularity and Reusability of LLMs through General Knowledge Subtraction

The rapid evolution of large language models (LLMs) has revolutionized natural language processing (NLP) tasks such as text generation, translation, and comprehension. However, the increasing computational demands and inference costs of these models present significant challenges. This study investigates the dynamic and efficient utilization of pre-trained weights from open-sourced LLMs of varying parameter sizes to achieve an optimal balance between computational efficiency and task performance. Drawing inspiration from the dual-process theory of human cognition, we introduce StitchLLM: a dynamic model routing framework that employs a powerful bottom model to process all queries, and uses a lightweight routing mechanism to allocate computational resources appropriately. Our novel framework optimizes efficiency and maintains performance, leveraging a trainable stitching layer for seamless integration of decoder layers across different LLMs. Experimental results demonstrate that StitchLLM improves system throughput while minimizing performance degradation, offering a flexible solution for deploying LLMs in resource-constrained settings.

StitchLLM: Serving LLMs, One Block at a Time

In this paper, we study format biases in reinforcement learning from human feedback (RLHF). We observe that many widely-used preference models—including human evaluators, GPT-4, and top-ranking models on the RewardBench benchmark—exhibit strong biases towards specific format patterns, such as lists, links, bold text, and emojis. Furthermore, large language models (LLMs) can exploit these biases to achieve higher rankings on popular benchmarks like AlpacaEval and LMSYS Chatbot Arena. One notable example is verbosity bias, where current preference models favor longer responses that appear more comprehensive, even when their quality is equal to or lower than shorter responses. However, format biases beyond verbosity remain largely underexplored. In this work, we extend the study of biases in preference learning beyond the commonly recognized length bias, offering a comprehensive analysis of a wider range of format biases. Additionally, we show that with a small amount of biased data (less than 1%), we can inject significant bias into the reward model. Moreover, these format biases can also be easily exploited by downstream alignment algorithms, such as *best-of-n sampling* and online iterative *DPO*, as it is usually easier to manipulate the format than to improve the quality of responses. Our findings emphasize the need to disentangle format and content both for designing alignment algorithms and evaluating models.

From Lists to Emojis: How Format Bias Affects Model Alignment

While recent research increasingly emphasizes the value of human-LLM collaboration in competitive programming and proposes numerous empirical methods, a comprehensive understanding remains elusive due to the fragmented nature of existing studies and their use of diverse, application-specific human feedback. Thus, our work serves a three-fold purpose: First, we present the first taxonomy of human feedback consolidating the entire programming process, which promotes fine-grained evaluation. Second, we introduce ELABORATIONSET, a novel programming dataset specifically designed for human-LLM collaboration, meticulously annotated to enable large-scale simulated human feedback and facilitate cost-effective real human interaction studies. Third, we introduce ELABORATION, a novel benchmark to facilitate a thorough assessment of human-LLM competitive programming. With ELABORATION, we pinpoint strengthes and weaknesses of existing methods, thereby setting the foundation for furture improvement. Our dataset and code will be openly released.

ELABORATION: A Comprehensive Benchmark on Human-LLM Competitive Programming

Grice's Quantity Maxims dictate that human speakers aim for the optimal quantity of information during conversation. To empower LLMs to self-repair their responses toward optimal quantity and improve their attentive listening skills, we propose Q-Tuning and Q-Traveling, which draw on heuristic path-finding to enable decoder-only LLMs to travel among multiple ``Q-alternatives'' (Quantity Alternatives) and search for the optimal quantity in coordination with a conversation goal. Automatic and human evaluations demonstrate the effectiveness of Q-Tuning and Q-Traveling in constructing human-like, user-centered conversation agents.

Towards LLM-powered Attentive Listener: A Pragmatic Approach through Quantity Self-Repair

The effective utilization of structured data, integral to corporate data strategies, has been challenged by the rise of large language models (LLMs) capable of processing unstructured information. This shift prompts the question: can LLMs interpret structured data directly in its unstructured form? We propose an automatic evaluation data generation method for assessing LLMs' reasoning capabilities on structure-rich text to explore this. Our approach supports 8 structured languages and 29 tasks, generating data with adjustable complexity through controllable nesting and structural width. We introduce StrucText-Eval, a benchmark containing 5,800 pre-generated and annotated samples designed to evaluate how well LLMs understand and reason through structured text. StrucText-Eval is divided into two suites: a regular Test suite (3,712 samples) and a Test-Hard suite (2,088 samples), the latter emphasizing the gap between human and model performance on more complex tasks. Experimental results show that while open-source LLMs achieve a maximum accuracy of 74.9% on the standard dataset, their performance drops significantly to 45.8% on the harder dataset. In contrast, human participants reach an accuracy of 92.6% on StrucText-Eval-Hard, highlighting LLMs' current limitations in handling intricate structural information.

Downloads

Next from ACL 2025

Serial Lifelong Editing via Mixture of Knowledge Experts

Similar lecture

Do LLMs Align Human Values Regarding Social Biases? Judging and Explaining Social Biases with LLMs