United States

Rapid improvements in large language models have unveiled a critical challenge in human-AI interaction: sycophancy. In this context, sycophancy refers to the tendency of models to excessively agree with or flatter users, often at the expense of factual accuracy. While previous studies have primarily analyzed this behavior in single-turn interactions, its persistence and evolution in multi-step conversations remain largely unexplored. We introduce METHODNAME , a benchmark specifically designed to evaluate sycophancy in extended dialogues, where language models must navigate iterative user feedback, challenges, and persuasion. We prompt models to elicit four types of sycophantic biases. We then propose and test sycophancy reduction strategies, evaluating their effectiveness beyond single-step interactions.

NAACL 2025

TRUTH DECAY: Quantifying Multi-Turn Sycophancy in Language Models

sycophancy

multi-turn

language model

bias

## 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.

poster

### Welcome to 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics

Welcome to the 2025 meeting of the Nations of the Americas Chapter of the Association for Computational Linguistics! I am proud to help organize the first NAACL conference to carry the new name of our organization, one that emphasizes inclusion for all of the Americas. I am also pleased to welcome you to Albuquerque, New Mexico, a state whose unique blend of cultural influences will make for an excellent backdrop for NAACL 2025, especially with this year’s special theme on NLP in a Multicultural World. 
**[Continue reading...](https://drive.google.com/file/d/1jX-qGhqVSZZCIrAnJaz798pflu5Irrdn/view?usp=sharing)**

*- Colin Cherry, Google, NAACL 2025 General Chair* 

[![](https://assets.underline.io/markdown_image/1/image/b087f8a4dc5816d6e1a6514e59c59ac3.png)](https://drive.google.com/file/d/1T96GzPqObXrMln2BMByCSXTSizjTg69P/view?usp=sharing)

You need to log in with the email address you registered with. Access credentials have been sent to your email. 

Please be sure to check your spam and other email folders if you do not see an email confirmation right away.

Please log in to explore this event.

To access NAACL 2025 event page you are required to register. Please follow [**this link**](https://2025.naacl.org/registration/registration/) to register. Access will depend on your registration type.

Please register!

Welcome to 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics

The recent large-scale emergence of LLMs has left an open space for dealing with their consequences, such as plagiarism or the spread of false information on the Internet. Coupling this with the rise of AI detector bypassing tools, reliable machine-generated text detection is in increasingly high demand. We investigate the paraphrasing attack resilience of various machine-generated text detection methods, evaluating three approaches: fine-tuned RoBERTa, Binoculars, and text feature analysis, along with their ensembles using Random Forest classifiers. We discovered that Binoculars-inclusive ensembles yield the strongest results, but they also suffer the most significant losses during attacks. In this paper, we present the dichotomy of performance versus resilience in the world of AI text detection, which complicates the current perception of reliability among state-of-the-art techniques.

Paraphrasing Attack Resilience of Various Machine-Generated Text Detection Methods

In recent years stylometry has been more investigated in many different fields. Hence, in this work, we are going to study detecting, generating, and evaluating the writing styles of different authors. In the first step, the sentences will be extracted from several different books, each belonging to a different author, to create a dataset. Then the selected models will be trained to detect the author of sentences in the dataset. After that generator models will be trained without paired dataset to generate sentences based on the authors' writing styles. Finally, in order to evaluate the performance of generators, the first trained models will be leveraged to evaluate the generated sentences.

Detecting, Generating, and Evaluating in the Writing Style of Different Authors

Data visualization is integral to any Exploratory Data Analysis (EDA) task. However, generating visualization requires expertise, presenting a steep learning curve and a significant cognitive load. Natural Language Interfaces (NLIs) for EDA aim to lower this barrier by allowing users to generate visualizations through natural language queries. However, complexity remains when EDA is performed collaboratively, requiring an environment to support multi-user interaction. In this thesis proposal, we discuss challenges in user-system interaction in a collaborative multi-user setup, such as errors in visualization generation due to misinterpretation of user requests. We hypothesize that a Conversational Assistant (CA) capable of understanding user-initiated clarification requests and generating accurate responses can improve user experience and support collaborative EDA tasks. To this end, we propose to develop such a CA (Figure [tab:system_issues]) and evaluate it through a user study, thus examining its impact on user experience in a collaborative environment for EDA.

Collaborative Data Exploration through Visualization: A Thesis Proposal Analyzing Impact of Conversational Assistants

This paper proposes SkipCLM, a novel method for improving multilingual machine translation in decoder transformers. We augment contrastive learning for cross-lingual alignment with a trainable skip connection to preserve information crucial for accurate target language generation. Experiments with XGLM-564M on the Flores-101 benchmark demonstrate improved performance, particularly for en-de and en-zh direction translation, compared to both direct sequence-to-sequence training and existing contrastive learning methods.

SkipCLM: Enchancing Crosslingual Alignment of Decoder Transformer Models via Contrastive Learning and Skip Connection

The proportion of responses to a question and its options, known as the response distribution, enables detailed analysis of human society. Recent studies highlight the use of Large Language Models (LLMs) for predicting response distributions as a cost-effective survey method. However, the reliability of these predictions remains unclear. LLMs often generate answers by blindly following instructions rather than applying rational reasoning based on pretraining-acquired knowledge. This study investigates whether LLMs can rationally estimate distributions when presented with explanations of ``artificially generated distributions'' that are against commonsense. Specifically, we assess whether LLMs recognize counterintuitive explanations and adjust their predictions or simply follow these inconsistent explanations. Results indicate that smaller or less human-optimized LLMs tend to follow explanations uncritically, while larger or more optimized models are better at resisting counterintuitive explanations by leveraging their pretraining-acquired knowledge. These findings shed light on factors influencing distribution prediction performance in LLMs and are crucial for developing reliable distribution predictions using language models.

Reliability of Distribution Predictions by LLMs: Insights from Counterintuitive Pseudo-Distributions

LLMs have demonstrated remarkable proficiency in understanding tasks but continue to struggle with long-context comprehension, particularly with content located in the middle of extensive inputs. This limitation, known as the Lost-in-the-Middle (LITM) problem, hinders models from fully processing and utilizing information across lengthy contexts. To address this issue, we introduce pause-tuning, a technique that redistributes attention to enhance comprehension of long-context inputs. Our approach involves fine-tuning language models on datasets with artificially inserted pause tokens, which serve to segment the input into smaller, more manageable parts. We evaluate pause-tuning against alternative approaches using the Needle-in-a-Haystack benchmark, where models must retrieve information embedded within contexts of up to 128K tokens. Experimental results demonstrate significant performance gains, with the LLaMA 3.2 3B Instruct model and the LLaMA 3.1 8B Instruct model improving by 10.61% and 3.57% respectively on average, suggesting that pause-tuning successfully enhances attention redistribution and improves long-context retention. The code and data are
available at <https://anonymous.4open.science/r/LITM-PauseTokens-7357>.

Pause-Tuning for Long-Context Comprehension: A Lightweight Approach to LLM Attention Recalibration

Theory of Mind (ToM), the ability to understand the mental states of oneself and others, remains a challenging area for large language models (LLMs), which often fail to predict human mental states accurately. In this paper, we introduce UniToMBench, a unified benchmark that integrates the strengths of SimToM and TOMBENCH to systematically improve and assess ToM capabilities in LLMs by integrating multi-interaction task designs and evolving story scenarios. Supported by a custom dataset of over 1,000 hand-written scenarios, UniToMBench combines perspective-taking techniques with diverse evaluation metrics to better stimulate social cognition in LLMs. Through evaluation, we observe that while models like GPT-4o and its experimental version show consistently high accuracy in tasks involving emotional and belief-related scenarios, with results usually above 80%, there is significant variability in their performance across knowledge-based tasks. These results highlight both the strengths and limitations of current LLMs in ToM-related tasks, underscoring the value of UniToMBench as a comprehensive tool for future development.

UniToMBench: Integrating Perspective-Taking to Improve Theory of Mind in LLMs

Large Language Models have demonstrated remarkable capabilities in natural language processing tasks, including mathematical problem-solving that requires multi-step logical reasoning. However, challenges persist in automating the identification of key mathematical concepts, understanding their interrelations, and formalizing proofs within a rigorous framework. We present a novel framework that leverages knowledge graphs to augment LLMs to construct and formalize mathematical proofs. Our results demonstrate significant performance improvements across multiple datasets, with using knowledge graphs, achieving up to a 34% success rate on the MUSTARDSAUCE dataset on o1-mini and consistently outperforming baseline approaches by 2-11% across different models. We show how this approach bridges the gap between natural language understanding and formal logic proof systems and achieve elevated results for foundation models over baseline.

Automating Mathematical Proof Generation Using Large Language Model Agents and Knowledge Graphs

There has been an increasing interest in the alignment of large language models (LLMs) with human values. However, the safety issues of their integration with a vision module, or vision language models (VLMs), remain relatively underexplored. In this paper, we propose a novel jailbreaking attack against VLMs, aiming to bypass their safety barrier when a user inputs harmful instructions. A scenario where our poisoned (image, text) data pairs are included in the training data is assumed. By replacing the original textual captions with malicious jailbreak prompts, our method can perform jailbreak attacks with the poisoned images. Moreover, we analyze the effect of poison ratios and positions of trainable parameters on our attack's success rate. For evaluation, we design two metrics to quantify the success rate and the stealthiness of our attack. Together with a list of curated harmful instructions, a benchmark for measuring attack efficacy is provided. We demonstrate the efficacy of our attack by comparing it with baseline methods.

ImgTrojan: Jailbreaking Vision-Language Models with ONE Image

The ability to revise essays in response to feedback is important for students' writing success. An automated writing evaluation (AWE) system that supports students in revising their essays is thus essential. We present eRevise+RF, an enhanced AWE system for assessing student essay revisions (e.g., changes made to an essay to improve its quality in response to essay feedback) and providing revision feedback. We deployed the system with 6 teachers and 406 students across 3 schools in Pennsylvania and Louisiana. The results confirmed its effectiveness in (1) assessing student essays in terms of evidence usage, (2) extracting evidence and reasoning revisions across essays, and (3) determining revision success in responding to feedback. The evaluation also suggested eRevise+RF is a helpful system for young students to improve their argumentative writing skills through revision and formative feedback.

Next from NAACL 2025

Paraphrasing Attack Resilience of Various Machine-Generated Text Detection Methods

Similar lecture

Self-Augmented Preference Alignment for Sycophancy Reduction in LLMs