Artificial intelligence (AI) is transforming elementary STEM education, yet evidence remains fragmented. This systematic review synthesizes 258 studies (2020-2025) examining AI applications across eight categories: intelligent tutoring systems (45% of studies), learning analytics (18%), automated assessment (12%), computer vision (8%), educational robotics (7%), multimodal sensing (6%), AI-enhanced extended reality (XR) (4%), and adaptive content generation. The analysis shows that most studies focus on upper elementary grades (65%) and mathematics (38%), with limited cross-disciplinary STEM integration (15%). While conversational AI demonstrates moderate effectiveness (d = 0.45-0.70 where reported), only 34% of studies include standardized effect sizes. Eight major gaps limit real-world impact: fragmented ecosystems, developmental inappropriateness, infrastructure barriers, lack of privacy frameworks, weak STEM integration, equity disparities, teacher marginalization, and narrow assessment scopes. Geographic distribution is also uneven, with 90% of studies originating from North America, East Asia, and Europe. Future directions call for interoperable architectures that support authentic STEM integration, grade-appropriate design, privacy-preserving analytics, and teacher-centered implementations that enhance rather than replace human expertise.

Evidence-based medicine (EBM) holds a crucial role in clinical application. Given suitable medical articles, doctors effectively reduce the incidence of misdiagnoses. Researchers find it efficient to use large language models (LLMs) techniques like RAG for EBM tasks. However, the EBM maintains stringent requirements for evidence, and RAG applications in EBM struggle to efficiently distinguish high-quality evidence. Therefore, inspired by the meta-analysis used in EBM, we provide a new method to re-rank and filter the medical evidence. This method presents multiple principles to filter the best evidence for LLMs to diagnose. We employ a combination of several EBM methods to emulate the meta-analysis, which includes reliability analysis, heterogeneity analysis, and extrapolation analysis. These processes allow the users to retrieve the best medical evidence for the LLMs. Ultimately, we evaluate these high-quality articles and show an accuracy improvement of up to 11.4% in our experiments and results. Our method successfully enables RAG to extract higher-quality and more reliable evidence from the PubMed dataset. This work can reduce the infusion of incorrect knowledge into responses and help users receive more effective replies.

Objective and scalable measurement of teaching quality is a persistent challenge in education. While Large Language Models (LLMs) offer potential, general-purpose models have struggled to reliably apply complex, authentic classroom observation instruments. This paper uses custom LLMs built on sentence-level embeddings, an architecture better suited for the long-form, interpretive nature of classroom transcripts than conventional subword tokenization. We systematically evaluate five different sentence embeddings under a data-efficient training regime designed to prevent overfitting. Our results demonstrate that these specialized models can achieve human-level and even super-human performance with expert human ratings above 0.65 and surpassing the average human-human rater correlation. Further, through analysis of annotation context windows, we find that more advanced models-those better aligned with human judgments-attribute a larger share of score variation to lesson-level features rather than isolated utterances, challenging the sufficiency of single-turn annotation paradigms. Finally, to assess external validity, we find that aggregate model scores align with teacher value-added measures, indicating they are capturing features relevant to student learning. However, this trend does not hold at the individual item level, suggesting that while the models learn useful signals, they have not yet achieved full generalization. This work establishes a viable and powerful new methodology for AI-driven instructional measurement, offering a path toward providing scalable, reliable, and valid feedback for educator development.

Despite significant advances in vision-language models (VLMs), most existing work follows an English-centric design process, limiting their effectiveness in multilingual settings. In this work, we provide a comprehensive empirical study analyzing the impact of several multilingual design choices, such as training data composition, encoder selection, and text backbones. The result is TowerVision, a family of open multilingual VLMs for both image-text and video-text tasks, built upon the multilingual text-only model Tower+. TowerVision achieves competitive performance on multiple multimodal multilingual benchmarks and shows particular strength in culturally grounded tasks and multimodal translation. By incorporating visual and cultural context during fine-tuning, our models surpass existing approaches trained on substantially larger datasets, as demonstrated on ALM-Bench and Multi30K (image tasks) and ViMUL-Bench (video tasks). Alongside the models, we release VisionBlocks, a high-quality, curated vision-language dataset. Our findings highlight that multilingual vision-language training data substantially improves cross-lingual generalization -- both from high-resource to underrepresented languages and vice versa -- and that instruction-tuned LLMs are not always the optimal initialization point. To support further research, we publicly release all models, data, and training recipes.

Free-form gesture understanding is highly appealing for human-computer interaction, as it liberates users from the constraints of predefined gesture categories. However, the sole existing solution GestureGPT suffers from limited recognition accuracy and slow response times. In this paper, we propose Gestura, an end-to-end system for free-form gesture understanding. Gestura harnesses a pre-trained Large Vision-Language Model (LVLM) to align the highly dynamic and diverse patterns of free-form gestures with high-level semantic concepts. To better capture subtle hand movements across different styles, we introduce a Landmark Processing Module that compensate for LVLMs' lack of fine-grained domain knowledge by embedding anatomical hand priors. Further, a Chain-of-Thought (CoT) reasoning strategy enables step-by-step semantic inference, transforming shallow knowledge into deep semantic understanding and significantly enhancing the model's ability to interpret ambiguous or unconventional gestures. Together, these components allow Gestura to achieve robust and adaptable free-form gesture comprehension. Additionally, we have developed the first open-source dataset for free-form gesture intention reasoning and understanding with over 300,000 annotated QA pairs.

Video generative models have recently achieved notable advancements in synthesis quality. However, generating complex motions remains a critical challenge, as existing models often struggle to produce natural, smooth, and contextually consistent movements. This gap between generated and real-world motions limits their practical applicability. To address this issue, we introduce RealDPO, a novel alignment paradigm that leverages real-world data as positive samples for preference learning, enabling more accurate motion synthesis. Unlike traditional supervised fine-tuning (SFT), which offers limited corrective feedback, RealDPO employs Direct Preference Optimization (DPO) with a tailored loss function to enhance motion realism. By contrasting real-world videos with erroneous model outputs, RealDPO enables iterative self-correction, progressively refining motion quality. To support post-training in complex motion synthesis, we propose RealAction-5K, a curated dataset of high-quality videos capturing human daily activities with rich and precise motion details. Extensive experiments demonstrate that RealDPO significantly improves video quality, text alignment, and motion realism compared to state-of-the-art models and existing preference optimization techniques.

During 2024 and 2025 the discussion about the theorem-proving capabilities of large language models started reporting interesting success stories, mostly to do with difficult exercises (such as problems from the International Mathematical Olympiad), but also with conjectures [Feldman & Karbasi, arXiv:2509.18383v1] formulated for the purpose of verifying whether the artificial intelligence could prove it. In this paper we report a theorem proving feat achieved by ChatGPT by using a protocol involving different prover and verifier instances of the gpt-5 model working collaboratively. To make sure that the produced proofs do not suffer from hallucinations, the final proof is formally verified by the lean proof assistant, and the conformance of premises and conclusion of the lean code is verified by a human. Our methodology is by no means complete or exact. It was nonetheless able to solve five out of six 2025 IMO problems, and close about a third of the sixty-six number theory conjectures in [Cohen, Journal of Integer Sequences, 2025].

Effective generalization in robotic manipulation requires representations that capture invariant patterns of interaction across environments and tasks. We present a self-supervised framework for learning hierarchical manipulation concepts that encode these invariant patterns through cross-modal sensory correlations and multi-level temporal abstractions without requiring human annotation. Our approach combines a cross-modal correlation network that identifies persistent patterns across sensory modalities with a multi-horizon predictor that organizes representations hierarchically across temporal scales. Manipulation concepts learned through this dual structure enable policies to focus on transferable relational patterns while maintaining awareness of both immediate actions and longer-term goals. Empirical evaluation across simulated benchmarks and real-world deployments demonstrates significant performance improvements with our concept-enhanced policies. Analysis reveals that the learned concepts resemble human-interpretable manipulation primitives despite receiving no semantic supervision. This work advances both the understanding of representation learning for manipulation and provides a practical approach to enhancing robotic performance in complex scenarios.

Although LLMs have demonstrated improved performance by scaling parallel test-time compute, doing so relies on generating reasoning paths that are both diverse and accurate. For challenging problems, the forking tokens that trigger diverse yet correct reasoning modes are typically deep in the sampling tree. Consequently, common strategies to encourage diversity, such as temperature scaling, encounter a worsened trade-off between diversity and accuracy. Motivated by this challenge, we treat parallel reasoning as a set-of-next-token-prediction problem, and incorporate a set-based global loss into Supervised Fine-Tuning (SFT) using self-supervised bipartite matching between our global forking tokens and unique reasoning traces. We observe that, while naive fine-tuning with multiple reasoning traces collapses these unique reasoning modes, our proposed method, Set Supervised Fine-Tuning (SSFT), preserves these modes and produces emergent global forking tokens. Experiments on multiple reasoning benchmarks show that our SSFT consistently outperforms SFT under both Pass@1 and Cons@k metrics.

To support latency-sensitive AI applications ranging from autonomous driving to industrial robot manipulation, 6G envisions distributed ML with computational resources in mobile, edge, and cloud connected over hyper-reliable low-latency communication (HRLLC). In this setting, speculative decoding can facilitate collaborative inference of models distributively deployed: a lightweight on-device model locally generates drafts while a more capable remote target model on a server verifies and corrects them in parallel with speculative sampling, thus resulting in lower latency without compromising accuracy. However, unlike autoregressive text generation, behavior cloning policies, typically used for embodied AI applications, cannot parallelize verification and correction for multiple drafts as each generated action depends on observation updated by a previous action. To this end, we propose Action Deviation-Aware Hybrid Inference (ADAHI), wherein drafts are selectively transmitted and verified based on action deviation, which has a strong correlation with action's rejection probability by the target model. By invoking server operation only when necessary, communication and computational overhead can be reduced while accuracy gain from speculative sampling is preserved. Experiments on our testbed show that ADAHI reduces transmission and server operations by approximately 40%, lowers end-to-end latency by 39.2%, and attains up to 97.2% of the task-success rate of baseline that invokes speculative sampling for every draft embedding vector.