On day three of the AI Engineer World's Fair, speakers emphasized the need for balance between AI automation and human agency, pushing back against fully automated 'software factory' visions.

Cotal is an open coordination layer for AI agents, addressing the high cost and failure rates caused by coordination breakdowns in multi-agent systems. It provides a shared space, identity verification, access control, and a replayable log for efficient agent collaboration.

Braza is a live AI tutor that teaches you how to use any app directly on your screen, in your own language.

This article provides a technical audit of AI text detection service Pangram, highlighting that its accuracy drops drastically on mixed human-AI text, false positive rates vary greatly by individual and genre, and the company’s incentives promote overconfidence in results, risking witch hunts and damaging trust.

The Wall Street Journal reports SpaceX showed investors a 'slimmer than iPhone' AI phone prototype before its June IPO, running a Qualcomm chip and xAI OS. Musk denies it as 'utterly false,' though he previously said the company would make a phone if necessary.

XR Blocks is a lightweight, cross-platform JavaScript library built on three.js for rapidly prototyping advanced XR and AI experiences. It features hand tracking, gesture recognition, world understanding, and Gemini AI integration, with a desktop simulator for development. Targeting Chrome v136+ with WebXR on Android XR, it is developer-friendly and cross-platform.

Margarita is a deterministic scripting language that extends Markdown for programmable agent workflows. With variables, includes, and loops, it lets developers build predictable LLM applications while maintaining control over context and costs.

The article explores the imminent arrival of AI companions and lovers, arguing that they will profoundly alter human relationships and self-perception. It discusses the development of 'agentic' AI, concerns about addiction and commercial manipulation, and the potential for AI to train relationship skills.

This paper presents an open-source simulation framework for systematic comparison of remote center of motion (RCM) modeling approaches and image-based visual servoing (IBVS) control architectures in laparoscopic robots. The framework integrates three RCM models and six IBVS architectures, revealing key structural sensitivities through case studies, including the impact of tangent-plane definition, constraint dimensionality, open- vs closed-loop enforcement, and robustness near kinematic singularities. All resources are released to support reproducible research.

This paper proposes a framework reinterpreting active learning budget regimes as shifts in the dominant generalization mechanism. By reinterpreting PAC-style risk components as dynamic interacting terms, it proves dominance shifts are structurally unavoidable, identifying three phases: data-driven, transition, and model-driven. Experiments show AL efficiency depends on alignment between strategy inductive bias and the active bottleneck, and self-supervised representation shifts transitions earlier, highlighting representation quality's role. The work provides a unified framework for transition-aware AL algorithms.

This paper proposes an imitation learning framework that uses a graph-based auxiliary network to encode crowd interactions and a trajectory-level objective to capture spatiotemporal dynamics, outperforming existing baselines on simulation and real-world data.

The study extracts 1,219 sustained deceleration events from 234 urban driving logs of Argoverse 2 dataset, encodes each event with 19 kinematic features, and discovers four stable modes via K-means clustering with bootstrap stability analysis: anticipatory soft (62.8%), reactive closing (30.6%), brake-like jerk (4.8%), and outlier (1.8%). Only pair age shows a medium effect (ε²=0.085); scene geometry and vulnerable-road-user proximity have negligible effects. An early-event classifier achieves macro-F1=0.758 at 1.0 s, with scene context contributing +0.059 F1 over kinematics alone. Modes are regime-invariant at medium speed (ARI=0.817) but regime-dependent at low speed (ARI=0.166).

An invariant extended Kalman filter (IEKF) is developed for state estimation of serial rigid manipulators with an arbitrary number of links, formulated entirely within the Lie group SE(3). The group-affine property of the kinematic equations makes the linearised error dynamics autonomous, so the Riccati equation governs the true error covariance rather than a local approximation. A physically separated noise model treats gyroscope and accelerometer channels independently. The filter is structured as a modular chain of per-link IEKFs with linear computational cost. Exponential ultimate boundedness in mean square is established via a Lie algebra Lyapunov function.

Researchers developed FLYNN, a recurrent neural network based on the complete brain connectome of the fruit fly, for vision-based navigation. Compared to traditional networks, FLYNN exhibits superior robustness to out-of-distribution data and sensory loss, remaining functional even under total vision loss.

This paper presents GPAC, a four-layer decentralized control architecture for multiple quadrotors transporting a cable-suspended payload without central coordination or explicit communication of cable states. Each quadrotor independently estimates its load share from local measurements, enabling implicit coordination. The system integrates geometric control, anti-swing regulation, wind rejection, adaptive mass estimation via concurrent learning, and a priority-ordered safety filter based on control barrier functions. High-fidelity simulations demonstrate a mean tracking RMSE of 33.8 cm with low computational cost.

Service robots searching for household objects rely on spatial priors to reduce search cost, yet object locations can vary with resident traits. This paper proposes PerSim, a rigidity-gated hybrid policy that combines a trait-conditioned prior with a population-frequency baseline, personalizing only when placement behavior is variable. Through a human-calibrated simulation pipeline and a unified human study (N=200), they show that personalization is favored primarily for low-rigidity objects, while the population-frequency baseline remains strong for universally placed items. Offline tests show improvement over nearest discrete configuration matching, and home digital twin experiments demonstrate reduced expected search cost.

Researchers introduce EmbodimentSemantic, a dataset and benchmark for spatial grounding in vision-language-action systems. It represents scenes as directed object-relation-object triplets and includes real-world and simulator-grounded data. Experiments show current models struggle with depth-aware and viewpoint-dependent spatial structures.

MG-SpaIR is a training-data-free framework that restores clean images from single observations corrupted by mixed degradations. It uses implicit neural representations with a multi-grade coarse-to-fine hierarchy and explicit sparse regularization to improve fidelity and suppress artifacts. Experiments show it outperforms Deep Image Prior.

Proposes a system that automatically generates realistic assembly sequences and trains real-time inspection models using synthetic data. It requires only CAD models and step descriptions, can be deployed within an hour, and achieves 92.4% accuracy on a real-world assembly case.

Researchers propose DeCoDe, a technique that decomposes few-shot image classification into pairwise comparisons, enabling off-the-shelf multimodal LLMs to become powerful few-shot classifiers without additional training. It outperforms state-of-the-art methods on twelve datasets, with code open-sourced.

This paper introduces SegFS, a dual-stream fast-slow framework for open-vocabulary video instance segmentation (OV-VIS). By using a slow, accurate path on sparse keyframes and a fast, lightweight path on subsequent frames, it achieves up to 14x lower latency than the mobile-oriented MOBIUS model while maintaining competitive accuracy.

This paper presents PixelEyes, a multi-turn visual reasoning agent that decouples reasoning from perception to address the repeated localization failures of multimodal LLMs. It introduces mask-guided visual search and semantic-region breadth-first search, constructs the PixelEyes-6K dataset and Pinpoint-Bench benchmark, and demonstrates significant headroom for existing models.

Urban-scale Visual Place Recognition (VPR) aims to identify the geographic location of a query image by matching it against a geo-tagged database. However, recent methods overlook a serious long-tailed problem in urban datasets, biasing models toward well-photographed areas while failing in sparsely covered regions. This paper systematically characterizes this imbalance and proposes Distribution-Aware Place Recognition (DAPR), a model-agnostic plug-in framework that rebalances gradient contributions across head and tail classes. Within classification-retrieval pipelines, DAPR applies a multi-scale distance search mechanism to compute per-class distributional compactness, providing complementary gains at the retrieval stage. On the SF-XL benchmark, DAPR outperforms the previous baseline by 18.3% on test set v1 and 6.7% on v2, and achieves consistent improvements across methods and benchmarks.

This paper proposes a training-free evidence-injection framework that systematically mitigates hallucinations in medical MLLMs by recalibrating visual perception and anchoring textual reasoning with ROI priors from MedSAM and anatomical coordinate mapping. Evaluations show up to ~6% improvement in close-ended accuracy and ~35% reduction in open-ended hallucinations.

Proposes DiSIINet, a unified model based on Denoising Diffusion Implicit Models that jointly performs medical image enhancement and segmentation, with a Symbiotic Information Interaction module enabling dynamic feature-level exchange, achieving significant improvements on multi-modal medical datasets.

A new method called Multi-Scale Layer Attention (MSLA) is proposed to improve Oracle Bone Inscription recognition. It explicitly models multi-scale and cross-layer feature interactions to capture fine-grained details. Experiments show it outperforms existing attention mechanisms while maintaining computational efficiency.

SLIM-RL is a new reinforcement learning method for diffusion large language models (dLLMs). It bounds the commit risk of each rollout step with a tau-budget decoder, reducing aggregate risk without trajectory reconstruction. Using a trace-free random-masking objective with variance reduction, it matches TraceRL's best MATH500 accuracy at 0.46x training samples on SDAR-4B, and improves over TraceRL on math and code benchmarks.

This study applies machine learning to analyze 619 Inka khipus, revealing three structural clusters via unsupervised clustering, achieving 86% F1 score in provenance classification, and identifying cord twist direction as the key discriminator for imperial style. It also uncovers colonial acquisition bias in museum collections and independently validates the moiety structure of Santa Valley khipus.

ALEE is a new evaluation framework that uses Abstract Meaning Representations (AMR) to generate English minimal pairs with controlled semantic shifts, paired with translations into target languages, enabling fine-grained diagnostics for embedding models in any language with English parallel data. A large-scale study across 275+ languages reveals significant performance disparities correlated with language prevalence and subword tokenization.

This paper systematically audits existing KB-VQA benchmarks, uncovering widespread issues such as missing or contradicted answers, ambiguous questions, and trivial visual scenes, which render accuracy a misleading metric. The authors propose an audit-and-repair protocol and a multi-entity augmentation protocol to address these flaws, and demonstrate that corrected evaluations yield significantly different model rankings.