Did the World Just Gain Hundreds of Millions of Alzheimer's Patients? A new definition of Alzheimer's includes people without symptoms. Experts are fiercely divided over whether that makes sense--and whether the new drugs really help. Many have heralded a golden age of Alzheimer's research, excited by the promise of new amyloid-targeting antibodies. Yet a large meta-study of such antibodies concluded that they probably result in "little to no difference" in the decline of memory and of cognitive ability. This July, some ten thousand people descended on London for the Alzheimer's Association International Conference. Another three thousand of us joined virtually. I'm an anthropologist, accustomed to conferences where croissants at coffee breaks count as a luxury. This one opened with six trumpeters dressed as Buckingham Palace guards, followed by dancers who discoed, vogued, and high-kicked through the history of British pop. After the first day's sessions, participants dispersed to parties; at one, singers in sequins belted out covers while someone played an illuminated, transparent violin. One night, the London Eye was lit purple to honor the people in attendance. The mood matched the spectacle. Joanne Pike, the Alzheimer's Association's president and C.E.O., announced that recent advances had brought the campaign against the disease to a "pivotal moment." Maria Carrillo, its chief science officer, hailed a groundswell of progress. That optimism followed decades of disappointment. Between 1974 and 2020, the U.S. government spent roughly twenty billion dollars on Alzheimer's research without seeing the approval of a single treatment designed to act on the disease's underlying biology, let alone stop or reverse its course. Today, Alzheimer's afflicts roughly one in nine Americans over sixty-five and, along with related dementias, has been estimated to cost the economy about three-quarters of a trillion dollars a year. In surveys, it ranks with cancer as the disease people fear most. It typically starts by robbing patients of recent memories--conversations they just had, where they put important objects--before corroding the basic functions of adult independence, from paying bills to managing medications. Eventually, it leaves patients unable to speak, move, swallow, or recognize close relatives.

Healthcare AI's next test is integration While advanced AI excels at processing clinical data, healthcare's true test lies in overcoming deeply fragmented administrative workflows. The entrance of major AI companies into healthcare is a meaningful and welcome development, accelerating the technical foundation available to the industry. Their models are increasingly capable of processing long clinical records, interpreting complex terminology, comparing documentation against evidence and generating coherent summaries from large volumes of information. For clinicians, operators, and administrative teams who spend significant time searching through fragmented data, these advances are helping reduce cognitive burden and make high-value information easier to access. But healthcare leaders should not confuse model capability with operational capability. Healthcare's administrative challenges are caused by fragmented information, fragmented workflows, and fragmented accountability, not a lack of information.

Speculative decoding accelerates language model inference by using a fast drafter to propose candidate tokens that are then verified by a larger target model. Existing theory largely studies the stochastic, distribution-preserving setting, where the goal is to exactly sample from the target distribution. In contrast, many practical systems use greedy decoding, relaxed acceptance rules, or tree-based candidate sets, where success is governed by local ranking and threshold events rather than exact distributional equality. We develop a theory for these regimes. We identify that many common acceptance criteria have rejection regions that can be characterized as lower level sets of the target distribution. For these, we characterize the exact KL divergence required for rejection yielding exact certificates and sharp margin-based bounds for strict greedy decoding, additive and multiplicative relaxed acceptance, top-(m) relaxed criteria, and entropy-thresholded acceptance. We then extend the framework to greedy tree decoding, deriving exact and margin-only certificates for when the target greedy token remains covered by the drafter's top-(m) candidates. Finally, we evaluate the resulting certificates on Qwen3 models, showing that relaxed and tree-based criteria substantially enlarge the region of certified acceptance, especially on decoding steps with low target model distribution margin. These results complement existing distribution-preserving analyses of speculative decoding by characterizing the deterministic local acceptance events common in practical inference systems.

Classification and Regression Trees (CART) constitute one of the most influential paradigms in statistical learning. Although a variety of impurity measures have been proposed for different statistical models, these criteria are typically introduced on a case-by-case basis and analyzed separately. In this paper, we study CART through the lens of Bregman divergences. This perspective places the classical least-squares criterion, Poisson deviance, Kullback-Leibler-type losses, and other impurity measures associated with exponential-family models within a common framework. As a result, key ingredients of the CART methodology -- including node representatives, impurity measures, and split selection rules -- can be expressed and analyzed through general properties of convex functions rather than through separate model-specific constructions. Beyond the algorithmic formulation, we investigate theoretical properties of Bregman-based CART procedures. In particular, we analyze how geometric properties of the generating convex function influence impurity reductions and stability of recursive partitions. We also establish consistency results within the proposed framework, providing a unified theoretical treatment for a broad family of CART type procedures. Our results provide a geometric interpretation of impurity-based tree construction and show that many classical CART impurity criteria admit a common interpretation within a Bregman framework.

T propose o address a no the vel challenge Agent-as-a-Judge of evaluating framew time-v ork. Our arying method and construct complex s answers, task-specific we judg answer of ten e a frontier g correctness ents based agentic and on a search source tree-structured systems attribution.

Recent advances in multi-modal models have demonstrated strong performance in tasks such as image generation and reasoning. However, applying these models to the fire domain remains challenging due to the lack of publicly available datasets with high-quality fire domain annotations. To address this gap, we introduce DetectiumFire, a large-scale, multi-modal dataset comprising of 22.5k high-resolution fire-related images and 2.5k real-world fire-related videos covering a wide range of fire types, environments, and risk levels. The data are annotated with both traditional computer vision labels (e.g., bounding boxes) and detailed textual prompts describing the scene, enabling applications such as synthetic data generation and fire risk reasoning. DetectiumFire offers clear advantages over existing benchmarks in scale, diversity, and data quality, significantly reducing redundancy and enhancing coverage of real-world scenarios. We validate the utility of DetectiumFire across multiple tasks, including object detection, diffusion-based image generation, and vision-language reasoning. Our results highlight the potential of this dataset to advance fire-related research and support the development of intelligent safety systems. We release DetectiumFire to promote broader exploration of fire understanding in the AI community.

Large language models (LLMs) excel at zero-shot inference but continue to struggle with complex, multi-step reasoning. Recent methods that augment LLMs with intermediate reasoning steps such as Chain of Thought (CoT) and Program of Thought (PoT) improve performance but often produce undesirable solutions, especially in algorithmic domains. We introduce Per-Instance Program Synthesis (PIPS), a method that generates and refines programs at the instance-level using structural feedback without relying on task-specific guidance or explicit test cases. To further improve performance, PIPS incorporates a confidence metric that dynamically chooses between direct inference and program synthesis on a per-instance basis. Experiments across three frontier LLMs and 30 benchmarks including all tasks of Big Bench Extra Hard (BBEH), visual question answering tasks, relational reasoning tasks, and mathematical reasoning tasks show that PIPS improves the absolute harmonic mean accuracy by up to 8.6% and 9.4% compared to PoT and CoT respectively, and reduces undesirable program generations by 65.1% on the algorithmic tasks compared to PoT with Gemini-2.0-Flash.

As multimodal large models (MLLMs) continue to advance across challenging tasks, a key question emerges: What essential capabilities are still missing? A critical aspect of human learning is continuous interaction with the environment - not limited to language, but also involving multimodal understanding and generation. To move closer to human-level intelligence, models must similarly support multi-turn, multimodal interaction. In particular, they should comprehend interleaved multimodal contexts and respond coherently in ongoing exchanges. In this work, we present an initial exploration through the INTERMT - the first preference dataset for multi-turn multimodal interaction, grounded in real human feedback.

Military weapon systems and command-and-control infrastructure augmented by artificial intelligence (AI) have seen rapid development and deployment in recent years. However, the sociotechnical impacts of AI on combat systems, military decision-making, and the norms of warfare have been understudied. We focus on a specific subset of lethal autonomous weapon systems (LAWS) that use AI for targeting or battlefield decisions. We refer to this subset as AI-powered lethal autonomous weapon systems (AI-LAWS) and argue that they introduce novel risks--including unanticipated escalation, poor reliability in unfamiliar environments, and erosion of human oversight--all of which threaten both military effectiveness and the openness of AI research. These risks cannot be addressed by high-level policy alone; effective regulation must be grounded in the technical behavior of AI models. We argue that AI researchers must be involved throughout the regulatory lifecycle. Thus, we propose a clear, behavior-based definition of AILAWS--systems that introduce unique risks through their use of modern AI--as a foundation for technically grounded regulation, given that existing frameworks do not distinguish them from conventional LAWS. Using this definition, we propose several technically-informed policy directions and invite greater participation from the AI research community in military AI policy discussions.

Recent progress in Large Vision-Language Models (LVLMs) has enabled promising applications in medical tasks, such as report generation and visual question answering. However, existing benchmarks focus mainly on the final diagnostic answer, offering limited insight into whether models engage in clinically meaningful reasoning. To address this, we present CheXStruct and CXReasonBench, a structured pipeline and benchmark built on the publicly available MIMIC-CXR-JPG dataset. CheXStruct automatically derives a sequence of intermediate reasoning steps directly from chest X-rays, such as segmenting anatomical regions, deriving anatomical landmarks and diagnostic measurements, computing diagnostic indices, and applying clinical thresholds. CXReasonBench leverages this pipeline to evaluate whether models can perform clinically valid reasoning steps and to what extent they can learn from structured guidance, enabling fine-grained and transparent assessment of diagnostic reasoning. The benchmark comprises 18,988 QA pairs across 12 diagnostic tasks and 1,200 cases, each paired with up to 4 visual inputs, and supports multi-path, multi-stage evaluation including visual grounding via anatomical region selection and diagnostic measurements. Even the strongest of 12 evaluated LVLMs struggle with structured reasoning and generalization, often failing to link abstract knowledge with anatomically grounded visual interpretation.