arXiv:2603.27006v1 Announce Type: cross Abstract: Large language models produce em dashes at varying rates, and the observation that some models "overuse" them has become one of the most widely discussed markers of AI-generated text. Yet no mechanistic account of this pattern exists, and the parallel observation that LLMs default to markdown-formatted output has never been connected to it. We propose that the em dash is markdown leaking into prose -- the smallest surviving unit of the structural orientation that LLMs acquire from markdown-saturated training corpora. We present a five-step genealogy connecting training data composition, structural internalization, the dual-register status of the em dash, and post-training amplification. We test this with a two-condition suppression experiment across twelve models from five providers (Anthropic, OpenAI, Meta, Google, DeepSeek): when models are instructed to avoid markdown formatting, overt features (headers, bullets, bold) are eliminated or nearly eliminated, but em dashes persist -- except in Meta's Llama models, which produce none at all. Em dash frequency and suppression resistance vary from 0.0 per 1,000 words (Llama) to 9.1 (GPT-4.1 under suppression), functioning as a signature of the specific fine-tuning procedure applied. A three-condition suppression gradient shows that even explicit em dash prohibition fails to eliminate the artifact in some models, and a base-vs-instruct comparison confirms that the latent tendency exists pre-RLHF. These findings connect two previously isolated online discourses and reframe em dash frequency as a diagnostic of fine-tuning methodology rather than a stylistic defect.

arXiv:2603.29466v1 Announce Type: cross Abstract: Existing methods for quantifying predictive uncertainty in neural networks are either computationally intractable for large language models or require access to training data that is typically unavailable. We derive a lightweight alternative through two approximations: a first-order Taylor expansion that expresses uncertainty in terms of the gradient of the prediction and the parameter covariance, and an isotropy assumption on the parameter covariance. Together, these yield epistemic uncertainty as the squared gradient norm and aleatoric uncertainty as the Bernoulli variance of the point prediction, from a single forward-backward pass through an unmodified pretrained model. We justify the isotropy assumption by showing that covariance estimates built from non-training data introduce structured distortions that isotropic covariance avoids, and that theoretical results on the spectral properties of large networks support the approximation at scale. Validation against reference Markov Chain Monte Carlo estimates on synthetic problems shows strong correspondence that improves with model size. We then use the estimates to investigate when each uncertainty type carries useful signal for predicting answer correctness in question answering with large language models, revealing a benchmark-dependent divergence: the combined estimate achieves the highest mean AUROC on TruthfulQA, where questions involve genuine conflict between plausible answers, but falls to near chance on TriviaQA's factual recall, suggesting that parameter-level uncertainty captures a fundamentally different signal than self-assessment methods.

arXiv:2603.29522v1 Announce Type: cross Abstract: Modern language models (LMs) must be trained on many orders of magnitude more words of training data than human children receive before they begin to produce useful behavior. Assessing the nature and origins of this "data gap" requires benchmarking LMs on human-scale datasets to understand how linguistic knowledge emerges from children's natural training data. Using transcripts from the BabyView dataset (videos from children ages 6-36 months), we investigate (1) scaling performance at child-scale data regimes, (2) variability in model performance across datasets from different children's experiences and linguistic predictors of dataset quality, and (3) relationships between model and child language learning outcomes. LMs trained on child data show acceptable scaling for grammar tasks, but lower scaling on semantic and world knowledge tasks than models trained on synthetic data; we also observe substantial variability on data from different children. Beyond dataset size, performance is most associated with a combination of distributional and interactional linguistic features, broadly consistent with what makes high-quality input for child language development. Finally, model likelihoods for individual words correlate with children's learning of those words, suggesting that properties of child-directed input may influence both model learning and human language development. Overall, understanding what properties make language data efficient for learning can enable more powerful small-scale language models while also shedding light on human language acquisition.

arXiv:2603.29552v1 Announce Type: cross Abstract: Multilingualism is incredibly common around the world, leading to many important theoretical and practical questions about how children learn multiple languages at once. For example, does multilingual acquisition lead to delays in learning? Are there better and worse ways to structure multilingual input? Many correlational studies address these questions, but it is surprisingly difficult to get definitive answers because children cannot be randomly assigned to be multilingual and data are typically not matched between languages. We use language model training as a method for simulating a variety of highly controlled exposure conditions, and create matched 100M-word mono- and bilingual datasets using synthetic data and machine translation. We train GPT-2 models on monolingual and bilingual data organized to reflect a range of exposure regimes, and evaluate their performance on perplexity, grammaticality, and semantic knowledge. Across model scales and measures, bilingual models perform similarly to monolingual models in one language, but show strong performance in the second language as well. These results suggest that there are no strong differences between different bilingual exposure regimes, and that bilingual input poses no in-principle challenges for agnostic statistical learners.

arXiv:2603.29893v1 Announce Type: cross Abstract: Healthcare conversational AI agents shouldn't be optimized only for clean benchmark accuracy in production-first regime; they must be optimized for the lived reality of patient conversations, where audio is imperfect, intent is indirect, language shifts mid-call, and compliance hinges on how guidance is delivered. We present a production-validated framework grounded in real-time signals from 115M+ live patient-AI interactions and clinician-led testing (7K+ licensed clinicians; 500K+ test calls). These in-the-wild cues -- paralinguistics, turn-taking dynamics, clarification triggers, escalation markers, multilingual continuity, and workflow confirmations -- reveal failure modes that curated data misses and provide actionable training and evaluation signals for safety and reliability. We further show why healthcare-grade safety cannot rely on a single LLM: long-horizon dialogue and limited attention demand redundancy via governed orchestration, independent checks, and verification. Many apparent "reasoning" errors originate upstream, motivating vertical integration across contextual ASR, clarification/repair, ambient speech handling, and latency-aware model/hardware choices. Treating interaction intelligence (tone, pacing, empathy, clarification, turn-taking) as first-class safety variables, we drive measurable gains in safety, documentation, task completion, and equity in building the safest generative AI solution for autonomous patient-facing care. Deployed across more than 10 million real patient calls, Polaris attains a clinical safety score of 99.9%, while significantly improving patient experience with average patient rating of 8.95 and reducing ASR errors by 50% over enterprise ASR. These results establish real-world interaction intelligence as a critical -- and previously underexplored -- determinant of safety and reliability in patient-facing clinical AI systems.

arXiv:2603.29997v1 Announce Type: cross Abstract: Analogical reasoning is a key driver of human generalization in problem-solving and argumentation. Yet, analogies between narrative structures remain challenging for machines. Cognitive engines for structural mapping are not directly applicable, as they assume pre-extracted entities, whereas LLMs' performance is sensitive to prompt format and the degree of surface similarity between narratives. This gap motivates a key question: What is the impact of enhancing structural mapping with LLM-derived abstractions on their analogical reasoning ability in narratives? To that end, we propose a modular framework named YARN (Yielding Abstractions for Reasoning in Narratives), which uses LLMs to decompose narratives into units, abstract these units, and then passes them to a mapping component that aligns elements across stories to perform analogical reasoning. We define and operationalize four levels of abstraction that capture both the general meaning of units and their roles in the story, grounded in prior work on framing. Our experiments reveal that abstractions consistently improve model performance, resulting in competitive or better performance than end-to-end LLM baselines. Closer error analysis reveals the remaining challenges in abstraction at the right level, in incorporating implicit causality, and an emerging categorization of analogical patterns in narratives. YARN enables systematic variation of experimental settings to analyze component contributions, and to support future work, we make the code for YARN openly available.

arXiv:2603.26948v2 Announce Type: replace Abstract: Existing approaches for predictive process monitoring are sub-symbolic, meaning that they learn correlations between descriptive features and a target feature fully based on data, e.g., predicting the surgical needs of a patient based on historical events and biometrics. However, such approaches fail to incorporate domain-specific process constraints (knowledge), e.g., surgery can only be planned if the patient was released more than a week ago, limiting the adherence to compliance and providing less accurate predictions. In this paper, we present a neuro-symbolic approach for predictive process monitoring, leveraging Logic Tensor Networks (LTNs) to inject process knowledge into predictive models. The proposed approach follows a structured pipeline consisting of four key stages: 1) feature extraction; 2) rule extraction; 3) knowledge base creation; and 4) knowledge injection. Our evaluation shows that, in addition to learning the process constraints, the neuro-symbolic model also achieves better performance, demonstrating higher compliance and improved accuracy compared to baseline approaches across all compliance-aware experiments.

arXiv:2512.19576v4 Announce Type: replace-cross Abstract: Attitude control is essential for many satellite missions. Classical controllers, however, are time-consuming to design and sensitive to model uncertainties and variations in operational boundary conditions. Deep Reinforcement Learning (DRL) offers a promising alternative by learning adaptive control strategies through autonomous interaction with a simulation environment. Overcoming the Sim2Real gap, which involves deploying an agent trained in simulation onto the real physical satellite, remains a significant challenge. In this work, we present the first successful in-orbit demonstration of an AI-based attitude controller for inertial pointing maneuvers. The controller was trained entirely in simulation and deployed to the InnoCube 3U nanosatellite, which was developed by the Julius-Maximilians-Universit\"at W\"urzburg in cooperation with the Technische Universit\"at Berlin, and launched in January 2025. We present the AI agent design, the methodology of the training procedure, the discrepancies between the simulation and the observed behavior of the real satellite, and a comparison of the AI-based attitude controller with the classical PD controller of InnoCube. Steady-state metrics confirm the robust performance of the AI-based controller during repeated in-orbit maneuvers.

arXiv:2601.19066v2 Announce Type: replace-cross Abstract: Bug Reproduction Tests (BRTs) have been used in many Automated Program Repair (APR) systems, primarily for validating promising fixes and aiding fix generation. In practice, when developers submit a patch, they often implement the BRT alongside the fix. Our experience deploying agentic APR reveals that developers similarly desire a BRT within AI-generated patches to increase their confidence. However, canonical APR systems tend to generate BRTs and fixes separately, and focus on producing only the fix in the final patch. In this paper, we study agentic APR in the context of cogeneration, where the APR agent is instructed to generate both a fix and a BRT in the same patch. We evaluate the effectiveness of different cogeneration strategies on 120 human-reported bugs at Google and characterize different cogeneration strategies by their influence on APR agent behavior. We develop and evaluate patch selectors that account for test change information to select patches with plausible fixes (and plausible BRTs). Finally, we analyze the root causes of failed cogeneration trajectories. Importantly, we show that cogeneration allows the APR agent to generate BRTs for at least as many bugs as a dedicated BRT agent, without compromising the generation rate of plausible fixes, thereby reducing engineering effort in maintaining and coordinating separate generation pipelines for fix and BRT at scale.

arXiv:2603.22519v2 Announce Type: replace-cross Abstract: Textual Large Language Models (LLMs) provide a simple and familiar interface: a string of text is used for both input and output. However, the information conveyed to an LLM often has a richer structure and semantics, which is not conveyed in a string. For example, most prompts contain both instructions ("Summarize this paper into a paragraph") and data (the paper to summarize), but these are usually not distinguished when passed to the model. This can lead to model confusion and security risks, such as prompt injection attacks. This work addresses this shortcoming by introducing an LLM-native mark-up language, LLMON (LLM Object Notation, pronounced "Lemon"), that enables the structure and semantic metadata of the text to be communicated in a natural way to an LLM. This information can then be used during model training, model prompting, and inference implementation, leading to improvements in model accuracy, safety, and security. This is analogous to how programming language types can be used for many purposes, such as static checking, code generation, dynamic checking, and IDE highlighting. We discuss the general design requirements of an LLM-native markup language, introduce the LLMON markup language and show how it meets these design requirements, describe how the information contained in a LLMON artifact can benefit model training and inference implementation, and provide some preliminary empirical evidence of its value for both of these use cases. We also discuss broader issues and research opportunities that are enabled with an LLM-native approach.

arXiv:2603.28522v2 Announce Type: replace-cross Abstract: We present LAD, a real-time language--action planner with an interruptible architecture that produces a motion plan in a single forward pass (~20 Hz) or generates textual reasoning alongside a motion plan (~10 Hz). LAD is fast enough for real-time closed-loop deployment, achieving ~3x lower latency than prior driving language models while setting a new learning-based state of the art on nuPlan Test14-Hard and InterPlan. We also introduce RAD, a rule-based planner designed to address structural limitations of PDM-Closed. RAD achieves state-of-the-art performance among rule-based planners on nuPlan Test14-Hard and InterPlan. Finally, we show that combining RAD and LAD enables hybrid planning that captures the strengths of both approaches. This hybrid system demonstrates that rules and learning provide complementary capabilities: rules support reliable maneuvering, while language enables adaptive and explainable decision-making.

Nature, Published online: 01 April 2026; doi:10.1038/s41586-026-10310-3

DNA damage burden and inadequate repair in CUX2+ cortical layer 2/3 excitatory neurons contributes to selective vulnerability in neuroinflammatory injury.

Nature, Published online: 01 April 2026; doi:10.1038/s41586-026-10313-0

Findings suggest that neural crest fate bias predominantly emerges within the neural tube, and that only a minor subset of delaminated progenitors retain multipotency to generate both sensory and sympathetic derivatives.

Nature, Published online: 01 April 2026; doi:10.1038/s41586-026-10299-9

N-desethyl-fluornitrazene is a µ-opioid receptor agonist derived from nitazenes that has supramaximal intrinsic efficacy that produces analgesia with minimal adverse effects in rodent models.

Nature, Published online: 01 April 2026; doi:10.1038/s41586-026-10359-0

LIGO–Virgo–KAGRA’s fourth Gravitational-Wave Transient Catalog shows evidence of a clear pair-instability gap in the distribution of binary black-hole secondary masses but is absent in the larger primary masses.

Nature, Published online: 01 April 2026; doi:10.1038/s41586-026-10290-4

The transcription factor ATF4 is shown to regulate double-stranded DNA repair within vulnerable CUX2+ upper-layer 2/3 cortical neurons, enabling their survival during development.

Nature, Published online: 01 April 2026; doi:10.1038/s41586-026-10295-z

AhR functions as a neuronal brake on axon regeneration, integrating environmental sensing, protein homeostasis and metabolic signalling to control the balance between stress adaptation and axonal repair.

Nature, Published online: 01 April 2026; doi:10.1038/s41586-026-10203-5

A study of reproducibility in a stratified random sample of 600 papers published from 2009 to 2018 in 62 journals spanning the social and behavioural sciences finds higher reproducibility among more recent papers and papers from journals that require data sharing.

Nature, Published online: 01 April 2026; doi:10.1038/s41586-026-10340-x

The seventh cholera pandemic, driven by El Tor Vibrio cholerae, evolved separately in Bangladesh and India, with the Ganges Basin acting as the global source, and its evolution has been shaped by phage defence system changes, influencing severity and transmission.

Nature, Published online: 01 April 2026; doi:10.1038/s41586-025-09844-9

When 100 social and behavioural science claims were examined, 34% of reanalyses closely matched the original results, with 74% reaching the same conclusion, revealing limited robustness of single-path analyses and the need to address analytical uncertainty.