While existing benchmarks probe the reasoning abilities of large language models (LLMs) across diverse domains, they predominantly assess passive reasoning, providing models with all the information needed to reach a solution. By contrast, active reasoning-where an LLM must interact with external systems to acquire missing evidence or data-has received little systematic attention. To address this shortfall, we present AR-Bench, a novel benchmark designed explicitly to evaluate an LLM's active reasoning skills. AR-Bench comprises three task families-detective cases, situation puzzles, and guessing numbers-that together simulate real-world, agentic scenarios and measure performance across commonsense, logical, and symbolic reasoning challenges. Empirical evaluation on AR-Bench demonstrates that contemporary LLMs exhibit pronounced difficulties with active reasoning: they frequently fail to acquire or leverage the information needed to solve tasks. This gap highlights a stark divergence between their passive and active reasoning abilities. Moreover, ablation studies indicate that even advanced strategies, such as tree-based searching or post-training approaches, yield only modest gains and fall short of the levels required for real-world deployment. Collectively, these findings highlight the critical need to advance methodology for active reasoning, e.g., incorporating interactive learning, real-time feedback loops, and environment-aware objectives for training. The benchmark is publicly available at: https://github.com/tmlr-group/AR-Bench.

The funeral director said "AI" as if it were a normal element of memorial services, like caskets or flowers. Of all places, I had not expected artificial intelligence to follow me into the small, windowless room of the mortuary. But here it was, ready to assist me in the task of making sense of death. It was already Wednesday, and I'd just learned that I had to write an obituary for my mother by Thursday afternoon if I wanted it to run in Sunday's paper. AI could help me do this.

Existing benchmarks have proven effective for assessing the performance of fully trained large language models. However, we find striking differences in the early training stages of small models, where benchmarks often fail to provide meaningful or discriminative signals. To explore how these differences arise, this competition tackles the challenge of designing scientific knowledge evaluation tasks specifically tailored for measuring early training progress of language models. Participants are invited to develop novel evaluation methodologies or adapt existing benchmarks to better capture performance differences among language models. To support this effort, we provide three pre-trained small models (0.5B, 1B, and 3B parameters), along with intermediate checkpoints sampled during training up to 200B tokens. All experiments and development work can be run on widely available free cloud-based GPU platforms, making participation accessible to researchers with limited computational resources. Submissions will be evaluated based on three criteria: the quality of the performance signal they produce, the consistency of model rankings at 1 trillion tokens of training, and their relevance to the scientific knowledge domain. By promoting the design of tailored evaluation strategies for early training, this competition aims to attract a broad range of participants from various disciplines, including those who may not be machine learning experts or have access to dedicated GPU resources. Ultimately, this initiative seeks to make foundational LLM research more systematic and benchmark-informed from the earliest phases of model development.

Artificial intelligence is poised to transform teaching and coaching practices,yet its optimal role alongside human expertise remains unclear.This study investigates human and AI collaboration in fitness education through a one year qualitative case study with a Pilates instructor.The researcher participated in the instructor classes and conducted biweekly semi structured interviews to explore how generative AI could be integrated into class planning and instruction.

The winners of three International Joint Conferences on Artificial Intelligence (IJCAI) awards have been announced. These three distinctions are: the Award for Research Excellence, the Computers and Thought Award and the John McCarthy Award. The Research Excellence award is given to a scientist who has carried out a program of research of consistently high quality throughout an entire career yielding several substantial results. The winner of the 2025 Award for Research Excellence is Rina Dechter, Distinguished Professor of Computer Science, University of California, Irvine, USA . Professor Dechter is recognized for her seminal contributions to the fields of constraint satisfaction and probabilistic inference, including novel algorithmic frameworks, modeling ideas, complexity analyses, and unifying principles.

The NLP research community has made publicly available numerous instruments for measuring representational harms caused by large language model (LLM)-based systems. These instruments have taken the form of datasets, metrics, tools, and more. In this paper, we examine the extent to which such instruments meet the needs of practitioners tasked with evaluating LLM-based systems. Via semi-structured interviews with 12 such practitioners, we find that practitioners are often unable to use publicly available instruments for measuring representational harms. We identify two types of challenges. In some cases, instruments are not useful because they do not meaningfully measure what practitioners seek to measure or are otherwise misaligned with practitioner needs. In other cases, instruments - even useful instruments - are not used by practitioners due to practical and institutional barriers impeding their uptake. Drawing on measurement theory and pragmatic measurement, we provide recommendations for addressing these challenges to better meet practitioner needs.

Considering the proliferation of responses to Piantadosi's original paper and the ongoing debate sparked by this special issue of the Italian Journal of Linguistics, it is clear that the discussion has touched a raw nerve in linguistic theorizing . In the original target paper (Chesi, this issue), I illustrated three prototypical (and in many respects, extreme) positions -- the computational, theoretical, and experimental perspectives -- without explicitly endorsing any of them. Instead, I attempted to highlight what I believe are the key weaknesses o f each of these prototypical stances, ultimately concluding that formal (i.e., ' generative ') linguistics -- more specifically, Minimalis m, my theoretical comfort zone -- must adopt practices and tools that are common in both computational and experimental fields . As noted by most respondents, the title and some of the more extreme statements were intended as mild provocations to draw attention to core issues affecting linguistic theorizing . M y position -- somehow obscured behind the ' three - body problem ' -- is that any relevant scientific progress is driven by theoretical insight, not by trawling using experimental or computational methods that are cost - inefficient, energy - intensive, and ultimately unsustainable . Moreover, in full agreement with most of the replies, I believe that the success of certain large language models (L L Ms), which are based on specific architectural assumptions, do es not constitute a refutation of the generative paradigm. On the contrary, it strongly supports several key intuitions that have emerged within the generative linguistic tradition (Rizzi this issue) . H owever, a concrete problem of ' incommensurability ' arises (Hao this issue), as differing methodologies and specialized jargon (Butt this issue) often result in circular, unresolved discussions .

The misuse of Large Language Models (LLMs) to infer emotions from text for malicious purposes, known as emotion inference attacks, poses a significant threat to user privacy. In this paper, we investigate the potential of Apple Intelligence's writing tools, integrated across iPhone, iPad, and MacBook, to mitigate these risks through text modifications such as rewriting and tone adjustment. By developing early novel datasets specifically for this purpose, we empirically assess how different text modifications influence LLM-based detection. This capability suggests strong potential for Apple Intelligence's writing tools as privacy-preserving mechanisms. Our findings lay the groundwork for future adaptive rewriting systems capable of dynamically neutralizing sensitive emotional content to enhance user privacy. To the best of our knowledge, this research provides the first empirical analysis of Apple Intelligence's text-modification tools within a privacy-preservation context with the broader goal of developing on-device, user-centric privacy-preserving mechanisms to protect against LLMs-based advanced inference attacks on deployed systems.

Visually impaired people could benefit from Visual Question Answering (VQA) systems to interpret text in their surroundings. However, current models often struggle with recognizing text in the photos taken by this population. Through in-depth interviews with visually impaired individuals, we identified common framing conventions that frequently result in misaligned text. Existing VQA benchmarks primarily feature well-oriented text captured by sighted users, under-representing these challenges. To address this gap, we introduce ROtated SAm-pling ( ROSA), a decoding strategy that enhances VQA performance in text-rich images with incorrectly oriented text. ROSA outperforms Greedy decoding by 11.7 absolute points in the best-performing model.

As Large Language Models are increasingly deployed in high-stakes domains, their ability to detect false assumptions and reason critically is crucial for ensuring reliable outputs. False-premise questions (FPQs) serve as an important evaluation method by exposing cases where flawed assumptions lead to incorrect responses. While existing benchmarks focus on single-hop FPQs, real-world reasoning often requires multi-hop inference, where models must verify consistency across multiple reasoning steps rather than relying on surface-level cues. To address this gap, we introduce MultiHoax, a benchmark for evaluating LLMs' ability to handle false premises in complex, multi-step reasoning tasks. Our dataset spans seven countries and ten diverse knowledge categories, using Wikipedia as the primary knowledge source to enable factual reasoning across regions. Experiments reveal that state-of-the-art LLMs struggle to detect false premises across different countries, knowledge categories, and multi-hop reasoning types, highlighting the need for improved false premise detection and more robust multi-hop reasoning capabilities in LLMs.