Named after the hundred-eyed watchman of Greek myth, Argus watches the education landscape: spotting new opportunities, pressure-testing the ventures we're building, and tracing every read back to the real-world signals behind it.
The evidence library: the raw signals the pipeline is watching across the education ecosystem. Every idea is built from these.
arXiv:2607.06721v1 Announce Type: new Abstract: Building upon found examples is a popular way people learn to code, especially in creative coding communities where sharing projects and remixing are common practices. But effectively doing so requires being able to 1) understand how existing code works, and 2) extend it by writing code that implements your own ideas, practices that can be challenging for new creative coders. We explored how to support these two processes through the design of Flowcode, a creative coding programming environment that integrates a flowchart for visualizing code structure and a chat interface tailored to support learning to code over vibe coding. We share how we iterated on the design of Flowcode over two studies with new creative coders, reflecting on the roles visualization and friction may play in enabling productive AI-use in computing education.
arXiv:2607.06687v1 Announce Type: new Abstract: AI-mediated communication is increasingly being utilized to help facilitate interactions; however, in privacy sensitive domains, an AI mediator has the additional challenge of considering how to preserve privacy. In these contexts, a mediator may redact or withhold information, raising questions about how users perceive these interventions and whether explanations of system behavior can improve trust. In this work, we investigate how explanations of redaction operations can affect user trust in AI-mediated communication. We devise a scenario where a validated system removes sensitive content from messages and generates explanations of varying detail to communicate its decisions to recipients. We then conduct a user study with 180 participants that studies how user trust and preferences vary for cases with different amounts of redacted content and different levels of explanation detail. Our results show that participants believed our syste
arXiv:2607.06681v1 Announce Type: new Abstract: Knowledge workers switch between applications thousands of times per day, spending nearly a tenth of the work year transitioning between digital applications in a process called digital fragmentation. Whether this fragmentation reflects who an employee is, where they work, or what kind of day they are having, has remained an open question. We analyzed 103 million application events recorded second-by-second from 1,017 employees across eight organizations that largely employ knowledge workers (e.g., law, financial services). Day-to-day variation in fragmentation within individual employees accounted for 44.6% of the variation in digital fragmentation, slightly exceeding stable individual differences between employees (35.8%), and far exceeding variation between organizations (19.6%). Fragmentation rose over the work week and reset after weekends and holidays. Higher-than-typical use of communication applications coincided with more fragmen
arXiv:2604.11840v3 Announce Type: replace-cross Abstract: Language models are increasingly used to simulate people: survey respondents, negotiators, stakeholders in policy exercises. In that role a model should reproduce how people plausibly behave, hesitating, conceding late, and settling for imperfect deals, rather than playing the best move. We call this the sampler role, in contrast to the solver role of finding the best move, and we test how the reasoning modes providers ship to strengthen models as solvers affect it. Our testbed is multi-party negotiation: five agents bargain over a regulation for fifteen turns, and unresolved issues are decided by an authority. Agents without a structured memory of the negotiation almost never reach agreement, whether reasoning is on or off: 314 of 315 such runs end with the authority deciding. What reasoning changes is how the failure looks. With reasoning enabled, one model family negotiates visibly, with varied moves, concessions in most runs
arXiv:2603.26772v2 Announce Type: replace-cross Abstract: Automated semantic annotation of broadcast television content presents distinctive challenges, combining structured audiovisual composition, domain-specific editorial patterns, and strict operational constraints. While multimodal large language models (MLLMs) have demonstrated strong general-purpose video understanding capabilities, their comparative effectiveness across pipeline architectures and input configurations in broadcast-specific settings remains empirically undercharacterized. This paper presents a systematic evaluation of multimodal annotation pipelines applied to broadcast television news in the Italian setting. We construct a domain-specific benchmark of clips labeled across four semantic dimensions: visual environment classification, topic classification, sensitive content detection, and named entity recognition. Two different pipeline architectures are evaluated across nine frontier models, including Gemini 3.0 P
arXiv:2509.08299v2 Announce Type: replace-cross Abstract: Scientific progress depends on researchers' ability to access and build upon the work of others. Yet, much published work remains behind expensive paywalls, and even accessible articles often rest on datasets shared only "upon reasonable request" to the authors. Researchers can try to overcome these barriers through informal channels, such as emailing authors directly, but whether such channels are hindered by racial or institutional biases remains unknown. Here we combine survey data, semi-structured interviews, large-scale observational analysis, and two randomized audit experiments to examine disparities in access to scientific knowledge. Surveyed researchers in the Global South report markedly lower institutional access to the literature and depend more heavily on informal channels to obtain papers and data; interviews elaborate the workarounds and racialized frictions they encounter. Our analysis of 250 million articles rev
arXiv:2402.01811v2 Announce Type: replace-cross Abstract: Credit scoring has been catalogued by the European Commission and the Executive Office of the US President as a high-risk classification task, in light of the potential harms of making loan approval decisions based on models that would be biased against certain groups. To address this concern, recent credit scoring research has considered a range of fairness-enhancing techniques put forward by the machine learning community to reduce bias and unfair treatment in classification systems. While the definition of fairness or the approach they follow to impose it may vary, most of these techniques, however, disregard the robustness of the results. This can create situations where unfair treatment is effectively corrected in the training set, but when producing out-of-distribution classifications, unfair treatment is incurred again. Instead, in this paper, we will investigate how to apply Distributionally Robust Optimisation (DRO) met
arXiv:2606.20995v2 Announce Type: replace Abstract: Artificial intelligence (AI) is reshaping scientific practices, yet its epistemic implications remain underanalyzed. While recent advances in large language models are substantial, machine learning (ML) has a deeper history across disciplines. This manuscript examines 4.9 million publications and 255 ML techniques to understand how the latter are reconfiguring scientific methods and knowledge production. Through embedding-based mapping, we reconstructed the semantic space of ML research, and found a core-periphery structure where physical sciences form the methodological core and health sciences represent the primary area of adoption. Methodological profiles vary by domain: predictive techniques are concentrated in computer sciences, while inferential approaches remain distributed across applied fields. Predictive architectures, however, are displacing inference-oriented techniques in domains that have traditionally prioritized interp
arXiv:2605.27395v2 Announce Type: replace Abstract: As the rapid proliferation of AI systems and harms spurs efforts in AI governance around the world, prioritizing among competing policy options has become increasingly challenging for policymakers and researchers. We introduce a methodology for identifying viable policy options to mitigate specified AI harms, helping policymakers and researchers target areas that warrant greater time and resource investment. This method combines participatory evaluation of policies, expert assessment of implementation costs, and an LLM-based assessment of perceived harm mitigation under each policy option. We leverage a genetic algorithm-based simulation study to explore a vast solution space of potential policy combinations, and examine how outcomes change under different weightings of cost, participatory input, and harm mitigation. We find that this method enables exploration of different balances between participatory and expert components, allowin
arXiv:2604.26075v2 Announce Type: replace Abstract: Governments increasingly deploy AI in public services, making transparency essential for accountability and public trust. Australia's Standard for AI Transparency Statements (AITS) requires government bodies to disclose how AI is used in practice, yet little empirical evidence exists on how these requirements are realised in documents. This paper presents a government AITS dataset, dubbed AITS-101, and provides one of the first systematic analysis of their content. Using stylometric, quantitative, and qualitative document analyses, we examine disclosure coverage, structure, and recurring patterns. Our findings reveal substantial variation in AI-related practice disclosure, highlight gaps between policy intent and implementation, and inform the design of more effective public-sector AI transparency standards.
arXiv:2603.03532v2 Announce Type: replace Abstract: Voting Advice Applications (VAA) are tools designed to help voters compare political candidates on policy preferences prior to elections. VAAs are popular tools in European countries and in other countries with multi-party democratic systems. Through a freedom of information request we got access to the inner workings of a popular Danish VAA called the 'textit{Kandidattest' which is implemented by a major Danish news outlet and has been used for general, municipal, and European elections. Users and politicians from every political party answer the same online questionnaire and get matched based on the agreement percentage stemming from their answers. VAAs play a significant role in elections with 45\% of surveyed voters reporting they followed their recommendations in the past Danish general election. However, the inner workings of VAAs have not been thoroughly evaluated until now. We find that the algorithm is not robust enough for u
arXiv:2601.18644v2 Announce Type: replace Abstract: The European Central Bank (ECB) is working on the "digital euro", an envisioned retail central bank digital currency for the Euro area. In this article, we take a closer look at the "digital euro FAQ", which provides answers to 26 frequently asked questions about the digital euro, and other published documents by the ECB on the topic. We question the provided answers based on our analysis of the current design in terms of privacy, technical feasibility, risks, costs and utility. In particular, we discuss the following key findings: (KF1) Central monitoring of all online digital euro transactions by the ECB threatens privacy even more than contemporary digital payment methods with segregated account databases. (KF2) The ECB's envisioned concept of a secure offline version of the digital euro offering full anonymity is in strong conflict with the actual history of hardware security breaches and mathematical evidence against it. (KF3) Th
arXiv:2508.15516v4 Announce Type: replace Abstract: Urban parks play a key role in supporting public health. Landscape architecture typically considers parks through the lens of form and function. While past research on equitable access has focused mainly on park form, studies addressing functional uses have been constrained by limited scale and coarse measurement techniques. Existing efforts have partially quantified park functions through small-scale surveys and movement data or general usage data, but have not effectively captured the specific activities and motivations underlying park visits. As a result, our understanding of the functional roles urban parks play remains incomplete. We introduce a novel method that refines mobile base station coverage using antenna azimuths, enabling more precise distinction of mobile traffic within parks versus surrounding areas. Using Paris as a case study, we analyze a large-scale dataset of passively collected per-app mobile network traffic acr
arXiv:2505.14215v3 Announce Type: replace Abstract: Despite the growing integration of retrieval-enabled AI agents into society, their safety and ethical behavior remain inadequately understood. In particular, the integration of LLMs and AI agents with external information sources and real-world environments raises critical questions about how they engage with and are influenced by these external data sources and interactive contexts. This study investigates how expanding retrieval access -- from no external sources to Wikipedia-based retrieval and open web search -- affects model reliability, bias propagation, and harmful content generation. Through extensive benchmarking of censored and uncensored LLMs and AI agents, our findings reveal a consistent degradation in refusal rates, bias sensitivity, and harmfulness safeguards as models gain broader access to external sources, culminating in a phenomenon we term safety degradation. Notably, retrieval-enabled agents built on aligned LLMs
arXiv:2607.07277v1 Announce Type: cross Abstract: Harmful online communication often contains slang, coded terms, abbreviations, and community-specific expressions, which make messages difficult to interpret. This paper presents an exploratory study of interpretation difficulty in Discord chats related to cybercrime. We construct reference interpretations of purposefully selected difficult messages, which were reviewed by an expert. We then use them to evaluate human and large language model (LLM) interpretations under different context conditions. The results show that local context alone is often insufficient for humans, while external knowledge and extended conversational context substantially improve human interpretation. For LLMs, local context also improves interpretation, and the larger model performs better. We further conduct a qualitative error analysis and propose a preliminary classification of factors that make harmful chats difficult to interpret. These findings suggest t
arXiv:2607.07652v1 Announce Type: new Abstract: Search engines have long allocated attention on the web by routing users from queries to websites. AI search changes this arrangement because information needs can be resolved inside the intermediary. Using URL-level Comscore U.S. desktop clickstream, we compare ChatGPT and Google information-seeking occasions and exploit ChatGPT Search access expansions to estimate traditional search displacement. ChatGPT produces outbound clicks in only 5.2% of conversation sessions, far below Google's referral ratio. The remaining clicks are not a scaled-down Google stream: they skew toward specialized destinations and away from ad-supported sites. Wider access cuts search use by 9.4%, with search-referral losses largest for informational categories. Our findings identify a central economic shift in digital intermediation: AI search might satisfy information needs inside the intermediary while weakening the referral bargain that has linked search, traf
arXiv:2607.07612v1 Announce Type: new Abstract: Artificial intelligence is rapidly evolving from generative systems to agentic AI capable of autonomously planning and executing tasks. Widely characterized as the Year of Agentic AI, 2025 marked accelerated development and deployment, introducing new ethical and governance challenges. This paper presents a systematic review of the emerging literature on agentic AI governance. Our analysis identifies features that distinguish agentic AI from traditional systems and why it warrants targeted governance attention. We synthesize prevailing governance priorities, proposed mechanisms, and stakeholder roles shaping this evolving domain. As an initial scholarly effort, this review lays the preliminary groundwork for developing a structured roadmap to guide responsible and adaptive agentic AI governance.
arXiv:2607.07605v1 Announce Type: new Abstract: This study adopts a behavioural bottom-up approach to AI value alignment to investigate whether an implicitly conveyed user identity shifts the moral evaluations of large language models (LLMs). Through a structured, multi-turn conversational protocol across 12,000 interactions, we evaluate AI value alignment in two non-reasoning models, gpt-4.1-mini-2025-04-14 and gemini-2.5-flash-lite. Rather than instructing the models to adopt a persona or prompting them with explicit moral stances, the user's professional role is introduced purely through value-neutral reasoning. The models are then asked for wrongness ratings from 0-100 on ten common-morality rules from Gert's moral framework. The results show that moral judgments vary with the user's role across both models. While grave-harm acts like killing exhibit a strong ceiling effect, contestable rule-governed acts demonstrate role-conditioned shifts that mirror the relationship between the
arXiv:2607.07267v1 Announce Type: new Abstract: Claims about the universality of human concepts have been predominantly assessed through linguistic similarity across languages and cultures. However, words are effective as communication devices because they compress rich experiential variation into shared conventions, potentially obscuring hidden individual and cultural differences in how concepts are mentally represented. Here, we analyse 2.6 billion human-made sketches of common concepts from 236 countries and territories to examine conceptual structure through people's visual imagination. Consistent with recent work on image-based cognition, we find that single concepts unfold into multiple distinct visual exemplars, revealing latent information about similarities and differences in conceptual structure across cultures. This variation is strongest for concepts involving haptic interaction, suggesting that visual imagery reflects variation in embodied experience as much as conventiona
arXiv:2607.06984v1 Announce Type: new Abstract: Misinformation often harms society not just by spreading a single false belief, but by breaking down the shared trust people rely on to evaluate what is true. This paper presents an agent-based simulation that frames trust as a collective resource and attention as a scarce private budget: when aggregate attention shifts toward low credibility content, the trust environment degrades, making credible information harder to process and correct. Across experiments, the model produces four recurring modes: credible stability, misinformation dominance, polarization, and a mixed baseline, with distinct signatures in trust trajectories and network structure. The results separate two control problems that matter for simulation-based policy exploration: the balance of trust repair versus harm largely determines whether the system recovers or collapses, while homophily and rewiring determine whether disagreement remains integrated or separates into p
arXiv:2607.06913v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly applied in public health applications, yet their robustness to non-clinical user inputs remains underexplored. We propose a domain specific robustness benchmark that evaluates LLMs under two perturbation types that commonly arise when non-clinical users interact with health AI systems: misinformation framing (MF), where prompt might be injected by false health claims, and layperson rewriting (LR), where patients describe symptoms in everyday language rather than medical terminology. Our goal is to evaluate the stability of LLMs under these perturbation. Experiments show that MF degrades accuracy by 7.2 pp on average with prediction flip rates of 9-38 percent, even when claims are explicitly labelled as unsupported; LR causes only 1.4 pp degradation. These findings highlight two distinct deployment risks in public health settings: models may produce incorrect outputs when users unintentionally
arXiv:2607.06784v1 Announce Type: new Abstract: In the era of digital revolution many contemporary events that changed the world were shaped through the internet. Nowadays, the emergence of internet of things (IoT), combining physical objects with virtual networks is expected to have even more influence. This new 'decentralised' structure in the world raises questions such as power, governance and the notion of democracy online. The aim of this paper is to investigate these notions. We have taken the examples of Bitcoin and Wikipedia and examined their decision-making process. Our analysis has found some inconsistencies in their policies, that are in contradiction with democracy and consensus principles of governance. Starting from our findings, we present further improvements that can be used to achieve more democracy and equity in the digital context.
Article URL: https://www.edalex.com/news/edalex-rich-skill-descriptor-rsd-library-openrsd-integrated-new-muzzy-lane-release-ai-driven-skillbuild-platform/ Comments URL: https://news.ycombinator.com/item?id=47698253 Points: 1 # Comments: 1
Three ways South Fayette Township School District brings their “Portrait of a Learner” to life.
Despite the lack of grammatical capitalisation, ispring is a really useful teaching tool.
Tech & Learning has partnered with the Ed-Tech Leadership Collective to explore how market pressures are affecting districts' vendor choices
K–12 school districts are packed with digital tools, and it’s up to IT leaders to manage the governance, data and risk associated with them. But no single department can do this alone and without all of the facts they need to make an informed decision. At ISTELive 2026 in Orlando, Fla., technology experts explained how cybersecurity, data privacy, accessibility and governance work together and why building a safe, intentional, student-centered digital environment depends on breaking down the silos between IT and the rest of the district. Cybersecurity Maturity Assessments Can Help Guide…
As the debate about screen time and digital tools in the classroom continues in school districts around the country, speakers at ISTELive 26 in Orlando, Fla., presented research noting that a balance between digital tools and foundational learning is the best path to effective learning. “This is not pro-tech versus anti-tech,” said Amanda Bollinger, associate administrator in the teaching and learning department at Jordan School District and a member of the Utah State Board of Education. “It’s just current reality. How do we balance those two things?” Cari Warnock, education…
arXiv:2606.13715v2 Announce Type: replace-cross Abstract: The best agent on WorkBench in March 2024, GPT-4, completed just 43% of tasks. We revisit the benchmark in June 2026 and find that the best agent to date, Claude Fable 5, now completes 98%. Beyond this considerable progress in frontier agent performance, three things stand out. First, unintended harmful actions, such as emailing the wrong person, fell from 26% of tasks for GPT-4 to 1.9% for Claude Fable 5; capability and safety go together on WorkBench rather than trade off, so the models that finish the most tasks also do the least unintended damage. Second, the rise of open-weight models has drastically lowered costs for a performance level that was only accessible to proprietary models, while frontier costs have stayed stable. Third, while several classes of error have been eliminated, frontier models still make some basic mistakes that occasionally result in irreversible harm. We release an updated version of the benchmark w
arXiv:2606.00684v2 Announce Type: replace-cross Abstract: We address the problem of out-of-distribution (OOD) detection for target observations embedded in a subspace of the high dimensional data space. Using continuous normalizing flows (CNFs), we propose a Lagrangian sub-flow (LSF) framework designed to isolate and estimate the density for the relevant components in the representation and using the remaining components as context. Through experimentation with models for speech synthesis, we show that CNFs, similarly to other deep generative models (DGMs), are susceptible to the "likelihood paradox", where high likelihood is erroneously assigned to OOD samples. This is attributed to the inductive bias of DGMs that prioritize low-level structural details over high-level semantic coherence. To mitigate this phenomenon, we propose a number of geometric diagnostic signals based on the velocity field over the sub-flow trajectory. Based on these signals, we design metrics for the challengin
arXiv:2604.22938v2 Announce Type: replace-cross Abstract: The promise of data-driven materials discovery remains constrained by the scarcity of large, high-quality, and accessible experimental datasets. Here, we introduce a generalizable large language model (LLM)-powered pipeline for automated extraction and structuring of materials data from unstructured scientific literature, using concrete materials as a representative and particularly challenging example. The pipeline exhibits robust performance across a broad range of LLMs and achieves an $F_1$ score of up to 0.98 for diverse composition--process--property attributes. Within one hour, it extracts nearly 9,000 high-quality records with over 100 attributes from a corpus screened from more than 27,000 publications, enabling the construction of the largest open laboratory database for blended cement concrete. Machine learning analyses underscore the importance of large, diverse, and information-rich datasets for enhancing both in-dis
arXiv:2603.19294v5 Announce Type: replace-cross Abstract: While post-training has successfully improved large language models (LLMs) across a variety of domains, these gains heavily rely on human-labeled data or external verifiers. Existing data has already been exploited, and new data is expensive to collect. Moreover, true intelligence goes far beyond verifiable tasks. Therefore, we need self-improvement frameworks that are less dependent on external signals and more broadly applicable to both verifiable and non-verifiable domains. We propose **Mutual Information Preference Optimization (MIPO)**, a contrastive data augmentation method that constructs preference pairs by generating a positive response conditioning on the correct prompt, and a negative response by conditioning on a random, unrelated prompt. We show that using Direct Preference Optimization to learn from this paired data maximizes pointwise mutual information *under the base LLM* between prompts and model responses. Exp
arXiv:2603.12056v3 Announce Type: replace-cross Abstract: Multimodal agents can now tackle complex reasoning tasks with diverse tools, yet they still suffer from inefficient tool use and inflexible orchestration in open-ended settings. A central challenge is enabling such agents to continually improve without parameter updates by learning from past trajectories. We identify two complementary forms of reusable knowledge essential for this goal: experiences, providing concise action-level guidance for tool selection and decision making, and skills, providing structured task-level guidance for planning and tool use. To this end, we propose XSkill, a dual-stream framework for continual learning from experience and skills in multimodal agents. XSkill grounds both knowledge extraction and retrieval in visual observations. During accumulation, XSkill distills and consolidates experiences and skills from multi-path rollouts via visually grounded summarization and cross-rollout critique. During
arXiv:2603.08316v3 Announce Type: replace-cross Abstract: Modern vision-language-model (VLM) based graphical user interface (GUI) agents are expected not only to execute actions accurately but also to respond to user instructions with low latency. While existing research on GUI-agent security mainly focuses on manipulating action correctness, the security risks related to response efficiency remain largely unexplored. In this paper, we introduce SlowBA, a novel backdoor attack that targets the responsiveness of VLM-based GUI agents. The key idea is to manipulate response latency by inducing excessively long reasoning chains under specific trigger patterns. To achieve this, we propose a two-stage reward-level backdoor injection (RBI) strategy that first aligns the long-response format and then learns trigger-aware activation through reinforcement learning. In addition, we design realistic pop-up windows as triggers that naturally appear in GUI environments, improving the stealthiness of
arXiv:2602.11103v2 Announce Type: replace-cross Abstract: Despite rapid progress on coding agents, progress on their multimodal counterparts has lagged behind. A key challenge is the scarcity of evaluation testbeds that combine the complexity of software development with the need for deep multimodal understanding. In game development, agents must navigate large, dense codebases while manipulating intrinsically multimodal assets such as shaders, sprites, and animations within a visual game scene. We present GameDevBench, the first benchmark for evaluating agents on game development tasks. GameDevBench consists of 333 tasks derived from web and video tutorials. Tasks require significant multimodal understanding and are complex: the average solution requires over three times the lines of code and file changes compared to prior software development benchmarks. Agents struggle with game development, with the best agent and method solving only 53.8% of tasks. We find a strong correlation bet
arXiv:2601.17020v2 Announce Type: replace-cross Abstract: With the rising popularity of interdisciplinary work and increasing institutional incentives in this direction, there is a growing need to understand how resulting publications incorporate ideas from multiple disciplines. Existing computational approaches, such as affiliation diversity, keywords, and citation patterns, do not account for how individual citations are used to advance the citing work. Although, in line with addressing this gap, prior studies have proposed taxonomies to classify citation purpose, these frameworks are not well-suited to interdisciplinary research and do not provide quantitative measures of citation engagement quality. To address these limitations, we propose a framework for the evaluation of citation engagement in interdisciplinary Natural Language Processing (NLP) publications. Our approach introduces a citation purpose taxonomy tailored to interdisciplinary work, supported by an annotation study. W
arXiv:2601.14660v2 Announce Type: replace-cross Abstract: Agentic Large Language Models (LLMs) are models able to reason, plan, and execute tools over unstructured data. These abilities are enabling transformative applications in domains spanning from personal assistant, financial, and legal domains. While these systems can substantially improve productivity and service quality, effective agency typically requires access to sensitive personal or organizational information. However, this access introduces critical inference-time privacy risks, specifically regarding contextually appropriate information disclosure. While recent studies highlight the inability of agentic LLMs to consistently adhere to privacy norms, existing defenses often rely on auxiliary LLM-based monitors. However, these defenses are expensive and offer limited protection against attacks that are robust to semantic censorship. To contrast this background, this paper proposes a notion of privacy filters based on activa
arXiv:2511.04500v3 Announce Type: replace-cross Abstract: Large language models (LLMs) are increasingly deployed as decision-making agents in high-stakes domains and as imitators of human behavior in the social and behavioral sciences. Yet how closely LLMs mirror human decision-making remains poorly understood. This gap is critical: misalignment could produce harmful outcomes in practice, while failure to replicate human behavior renders LLMs ineffective as social simulators. Here, we address this gap by replicating large-scale game-theoretic experiments and by introducing a systematic prompting and probing framework for machine-behavioral evaluation. We test three open models typically used to power agents (Llama, Mistral, and Qwen). Across 121 dyadic games spanning four classical game types, Llama reproduces human cooperation patterns with high fidelity, while Qwen aligns closely with Nash equilibrium predictions. Characterizing models through behavioral phenotyping, we find that hum
arXiv:2510.13537v2 Announce Type: replace-cross Abstract: On-device deployment of Large Language Models (LLMs) frequently leverages Low-Rank Adapters (LoRAs) to support diverse downstream tasks under tight resource constraints. To address the limited storage capacity of mobile devices, recent works have explored model merging techniques to fuse multiple LoRAs into a single one. In practice, however, LoRAs are often delivered incrementally, as users request support for new tasks (e.g., novel problem types or languages). This scenario introduces a new challenge: on-device online continual merging, where the objective is to incorporate new LoRAs while preserving the performance on previously supported tasks. In this paper, we propose a data-free and computationally efficient strategy for selecting and merging LoRAs when a new one becomes available, assuming the device can store only a limited number of adapters. Extensive experiments across real-world tasks demonstrate the superiority of
arXiv:2510.04140v2 Announce Type: replace-cross Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has become a widely adopted technique for enhancing the reasoning ability of Large Language Models (LLMs). However, the effectiveness of RLVR strongly depends on the capability of base models. This issue arises because it requires the model to have sufficient capability to perform high-quality exploration, which involves both effectiveness and diversity. Unfortunately, existing methods address this issue by imitating expert trajectories, which improve effectiveness but neglect diversity. To address this, we argue that the expert only needs to provide guidance only at critical decision points rather than the entire reasoning path. Based on this insight, we propose MENTOR: Mixed-policy Expert Navigation for Token-level Optimization of Reasoning, a framework that provides expert guidance only at critical decision points to perform effective and diverse exploration in RLVR. Exten
arXiv:2606.10531v2 Announce Type: replace Abstract: Quantization-aware training (QAT) is essential for extremely low-bit large language models (LLMs). Current QAT methods are mainly based on scalar quantization (SQ), which enables efficient optimization but suffers from severe performance degradation at 2-bit precision. On the other hand, vector quantization (VQ) provides substantially higher representational capacity, but its discrete codebook lookup prevents end-to-end training. We propose LC-QAT, a 2-bit weight-only VQ-QAT framework that represents quantized weights via a learned affine mapping over discrete vectors, which yields a high-quality PTQ initialization and enables fully differentiable end-to-end optimization without explicit codebook lookup in the training forward pass. This strong post-training initialization makes LC-QAT highly data-efficient. Experiments across diverse LLMs demonstrate that LC-QAT consistently outperforms state-of-the-art QAT methods while using only 0
arXiv:2606.10520v2 Announce Type: replace Abstract: Post-training quantization at the 2-bit level enables low-cost deployment and inference acceleration for large language models (LLMs). Scalar quantization (SQ) and vector quantization (VQ) are two primary quantization methods, however, the former suffers from significant performance degradation, and the latter incurs computational and storage overhead. We propose UniSVQ, a unified 2-bit quantization framework that bridges scalar and vector quantization by parameterizing codewords as an affine transform of integer lattices. This structure preserves compatibility with optimized integer kernels while retaining much of VQ's flexibility. We further introduce a data-driven block-wise fine-tuning strategy to directly minimize quantization reconstruction error. Extensive experiments across multiple LLM families and zero-shot benchmarks demonstrate that UniSVQ consistently outperforms state-of-the-art SQ methods and achieves performance compar
arXiv:2606.08625v2 Announce Type: replace Abstract: As Large Language Models (LLMs) advance toward open-ended autonomous agents, the mechanisms used to evaluate and guide their behavior must evolve accordingly. This work introduces the rubric as a unifying framework capturing this evolution, characterizing rubrics as a dynamic response to successive LLM paradigm shifts that recurs across otherwise independent efforts in evaluation, reinforcement learning, and safety alignment. We define rubrics as explicit criteria sets that transform complex quality judgments into structured and actionable standards, and demonstrate that their recurrence across these research threads is not coincidental. We systematically organize existing rubric designs, examine their construction and optimization, and analyze their role across evaluation and training. Rubrics manifest at three progressively deeper levels: at the evaluative level, they decompose holistic judgments into verifiable dimensions; at the t
arXiv:2605.03301v2 Announce Type: replace Abstract: De-identification of clinical text is a prerequisite for the secondary use of electronic health records. Existing public benchmarks such as the i2b2 2006 and 2014 corpora are over a decade old and lack the semantic and demographic diversity of modern clinical narratives. Large Language Models (LLMs) reach state-of-the-art zero-shot extraction, but their use at enterprise scale is limited by computational cost and by hospital data governance that restricts sending Protected Health Information (PHI) to cloud APIs. We introduce SHIELD (Synthetic Human-annotated Identifier-replaced Entries for Learning and De-identification), a diverse clinical note dataset of 1,381 notes with 10,229 gold-standard PHI spans across 9 categories, built with set-cover diversity sampling across demographic and document-type strata and human-in-the-loop adjudication. We evaluate four LLMs (two proprietary, two open-weight) to establish a performance ceiling on
arXiv:2604.12243v2 Announce Type: replace Abstract: Identifying promising research directions in fast-moving subareas is one of the most cognitively expensive tasks in modern AI research. Existing LLM-driven scientific discovery systems are typically limited to one-shot prompting on static literature snapshots and are validated only against contemporary judges such as human reviewers, agent peer review, wet-lab assays, or self-evaluation, leaving open whether they can anticipate future trends. We present Continuous Knowledge Metabolism (CKM), an AI workflow for hypothesis generation with three key capabilities: (i) continuous literature metabolism via sliding windows that maintain an evolving knowledge state; (ii) predictive evaluation, which grades hypotheses against papers published after the generation window; and (iii) practitioner-grade failure detection that diagnoses workflow failure modes from its outputs. On a 50-topic machine learning benchmark, CKM-Lite produces at least one
arXiv:2604.06817v2 Announce Type: replace Abstract: We present SemEval-2026 Task 9, a shared task on online polarization detection, covering 22 languages and comprising over 110K annotated instances. Each data instance is multi-labeled with the presence of polarization, polarization type, and polarization manifestation. Participants were asked to predict labels in three sub-tasks: (1) detecting the presence of polarization, (2) identifying the type of polarization, and (3) recognizing the polarization manifestation. The three tasks attracted over 1,000 participants worldwide and more than 10k submission on Codabench. We received final submissions from 67 teams and 73 system description papers. We report the baseline results and analyze the performance of the best-performing systems, highlighting the most common approaches and the most effective methods across different subtasks and languages. The dataset of this task is publicly available.
arXiv:2602.05711v2 Announce Type: replace Abstract: Mixture-of-Experts (MoE) architectures are evolving towards finer granularity to improve parameter efficiency. However, existing MoE designs face an inherent trade-off between the granularity of expert specialization and hardware execution efficiency. We propose OmniMoE, a system-algorithm co-designed framework that pushes expert granularity to its logical extreme. OmniMoE introduces vector-level Atomic Experts, enabling scalable routing and execution within a single MoE layer, while retaining a shared dense MLP branch for general-purpose processing. Although this atomic design maximizes capacity, it poses severe challenges for routing complexity and memory access. To address these, OmniMoE adopts a system-algorithm co-design: (i) a Cartesian Product Router that decomposes the massive index space to reduce routing complexity from O(N) to O(sqrt(N)); and (ii) Expert-Centric Scheduling that inverts the execution order to turn scattered,
arXiv:2601.04424v2 Announce Type: replace Abstract: Large language models (LLMs) now support contexts of up to 1M tokens, but their strengths and weaknesses on complex long-context tasks remain unclear. To study this, we focus on multi-document legal case summarization, where a single case often spans many documents exceeding 100K tokens. We systematically evaluate 12 frontier LLMs with Gavel, which consists of Gavel-Ref, a reference-based evaluation framework with checklist, residual-fact, and writing-style evaluations, and Gavel-Agent, a reference-free agent for evaluating factual coverage directly from source documents. Our results show that current models are more prone to omitting key information than hallucinating. They all perform well on simple checklist items, such as filing date, but struggle with rare and complex items, such as settlements. Performance also declines as case length increases. To meta-evaluate Gavel, we collect 160 hours of human annotations. Gavel-Agent reduc
arXiv:2511.14642v2 Announce Type: replace Abstract: Like visual processing, language processing is susceptible to illusions in which people systematically misperceive stimuli. In one such case--the comparative illusion (CI), e.g., More students have been to Russia than I have--comprehenders tend to judge the sentence as acceptable despite its underlying nonsensical comparison. Prior research has argued that this phenomenon can be explained as Bayesian inference over a noisy channel: the posterior probability of an interpretation of a sentence is proportional to both the prior probability of that interpretation and the likelihood of corruption into the observed (CI) sentence. Initial behavioral work has supported this claim by evaluating a narrow set of alternative interpretations of CI sentences and showing that comprehenders favor interpretations that are more likely to have been corrupted into the illusory sentence. In this study, we replicate and go substantially beyond this earlier
arXiv:2511.04694v5 Announce Type: replace Abstract: As large language model (LLM) based systems take on high-stakes roles in real-world decision-making, they must reconcile competing instructions from multiple sources within a single prompt context. Enforcing an instruction hierarchy, where higher-level directives override lower-priority requests, is critical to the reliability and control of LLMs. In this work, we reframe instruction hierarchy resolution as a reasoning task. The model must first "think" about the relationship between a given user prompt and higher-priority instructions before generating a response. To enable this capability, we construct VerIH, a training dataset of constraint-following tasks with verifiable answers, comprising aligned and conflicting system-user instructions. We show that lightweight reinforcement learning with VerIH effectively transfers general reasoning capabilities of models to instruction prioritization. Our method leads to consistent improvemen