Skip to content
AI News HubLIVE

This edition’s highlights

Policy

LifeVal

  • LifeVal draws on time-use studies and a 3,995-person representative survey to identify the everyday tasks people most want AI help with.
  • The benchmark covers caregiving, health, education, household administration, personal finances, shopping, and government services, emphasizing completed actions over chat advice.
OpenAI NewsIn-site articleLifeVal

Beyond Plausible Prediction: Safety Sufficiency in Web World Models

  • 86.1% of predictions were judged semantically plausible, but several models showed highly asymmetric authorisation behaviour.
  • Safety sufficiency asks whether a predictive interface preserves and exposes policy-decisive information for EXECUTE, ASK, or BLOCK decisions.
arXiv AIIn-site articleBeyond Plausible Prediction: Safety Sufficiency in Web World Models
Models

Import AI 479: Spacewar; low-latency tasks versus LLMs; Genie 4

  • Realtime-Bench reveals that LLMs perform poorly on tasks requiring real-time decision-making, with the best model achieving only 36% pass rate.
  • A DIA report details offensive space capabilities of China and Russia, including anti-satellite weapons and space-based nuclear threats.
Import AIIn-site articleImport AI 479: Spacewar; low-latency tasks versus LLMs; Genie 4

CaPtain: A Pipeline for Post-Training Code-as-Policy Robot Agents

  • Open post-training pipeline converts a small open VLM into a CaP robot agent without VLAs or frontier models.
  • Physical critic leverages privileged simulator state to credit multi-turn requests with delayed effects.
arXiv RoboticsIn-site articleCaPtain: A Pipeline for Post-Training Code-as-Policy Robot Agents

Drone-PUR: From Visual Clue Discovery to Understanding and Reasoning in UAV Scenes

  • Drone-PUR offers 11 tasks with 175k training and 20k test samples spanning single images, image sequences, cross-modal images, and multiple viewpoints.
  • Evaluating 22 general-purpose VLMs reveals persistent limitations in spatial judgment and cross-modal and cross-view reasoning.
arXiv Computer VisionIn-site articleDrone-PUR: From Visual Clue Discovery to Understanding and Reasoning in UAV Scenes

Do Open Language Models Invent Scripture? BILAL-BENCH: Benchmarking Islamic-Source Fabrication in Bengali

  • BILAL-BENCH contains 240 expert-verified questions covering Qur'anic attribution, hadith, fiqh, contemporary issues, and adversarial false-premise prompts, scored on two axes separating source errors from assertiveness.
  • Across 1,680 double-annotated responses, six of seven open-weight models confidently fabricate on 79.3–95.3% of source-production questions, and performance does not improve monotonically with scale.
arXiv Computational LinguisticsIn-site articleDo Open Language Models Invent Scripture? BILAL-BENCH: Benchmarking Islamic-Source Fabrication in Bengali
Agents

Towards Portable Specification of the Agent Harness

  • The harness is decomposed into eight functional components that can be specified and compared across systems
  • Different harness configurations can produce several-fold differences in reported scores for the same model
arXiv AIIn-site articleTowards Portable Specification of the Agent Harness

Qodo builds context into AI code review with Codex

  • Qodo's Codex plugin lets AI retrieve codebase rules and PR history before coding, and performs independent review after changes.
  • The plugin handles dependencies across 10,000+ repositories, connecting different Git providers or Kafka services.
OpenAI NewsIn-site articleQodo builds context into AI code review with Codex

Online Dense Reinforcement Learning with Planning-Grounding Agent for UI Navigation on Live iOS Apps

  • Online RL framework trains UI navigation agents directly on live iOS apps, avoiding brittle static trajectory memorization.
  • Decouples agent into planning agent and UI grounding model, with LLM-as-a-Judge providing dense step-level and terminal rewards.
Apple Machine Learning ResearchIn-site articleOnline Dense Reinforcement Learning with Planning-Grounding Agent for UI Navigation on Live iOS Apps

You have reached the end of this edition.

Your reading list
Other updates (50)
Agents

Scaling Git at OpenAI

OpenAI built Pando, a Git-compatible backend for large-scale concurrent development, reducing median push latency by nearly 5x and serving nearly a petabyte of Git data daily.

OpenAI NewsIn-site articleScaling Git at OpenAI

Lakebase Postgres is now generally available on Google Cloud

Databricks announces Lakebase Postgres is generally available on Google Cloud, offering serverless autoscaling, instant branching, and unified governance to help teams build apps and AI agents faster.

Databricks BlogIn-site articleLakebase Postgres is now generally available on Google Cloud

The U.S. Census Bureau and “Super Intelligence”

The U.S. Census Bureau’s Business Trends and Outlook Survey (BTOS) changed its AI question wording in November 2025 from “producing goods or services” to “any business function,” making pre- and post-change data incomparable. A later executive order, EO 14434, replaced “Artificial Intelligence (AI)” with “Super Intelligence (SI)” starting October 19, 2026, further complicating the survey series.

Simon Willison's WeblogIn-site articleThe U.S. Census Bureau and “Super Intelligence”

Agentic World Modeling for Agent Test-time Scaling without Environment Replica

The paper introduces Agentic World Modeling (AWM), which recasts next-observation prediction as a multi-turn evidence-acquisition task: an agentic simulator queries the real environment through safe read-only interfaces and constructs observations via executable computation without mutating shared state, enabling parallel trajectory exploration over one shared environment. It outperforms a single-shot simulation baseline by 40.07% and a trained world-model baseline by 12.36% on average across three benchmarks.

arXiv AIIn-site articleAgentic World Modeling for Agent Test-time Scaling without Environment Replica

Does Recursive Knowledge Construction Help Agents Verify Consensus Claims in Science?

A new paper investigates how recursive knowledge construction affects scientific agents' ability to verify consensus claims. The authors built ProClaim-eval, a dataset of 718 claims, and ProClaim, a system that persists an evidence state outside the model's context. They find that structured evidence state improves performance over in-context accumulation, and a trained judge outperforms fixed-round stopping. ProClaim ranks among the top three systems and matches Claude Sonnet 4.6 on three of five subsets.

arXiv AIIn-site articleDoes Recursive Knowledge Construction Help Agents Verify Consensus Claims in Science?

More Rules, Longer Lists? Benchmarking Self-Evolving Agents on a Hidden-Rule Curriculum

Researchers introduce CLR-Bench, a curriculum of latent rules that tests whether self-evolving agents can learn an unstated rule from feedback rather than simply appending exceptions to an ever-longer list. On the certifying model, six self-evolution methods improve little; with the stronger GPT-6-Sol, ACE and Dynamic Cheatsheet learn from feedback alone and outperform raw-history baselines.

arXiv AIIn-site articleMore Rules, Longer Lists? Benchmarking Self-Evolving Agents on a Hidden-Rule Curriculum

Coupled or Independent? Fault Coupling in Agentic Medical Vision Systems

This arXiv paper examines whether tool and reasoning channels in a dermatology vision-language agent fail together when the sensor reading itself is untrusted, showing that typographic injection causes detectable directional retrieval capture, and that hard-label channel comparison cannot detect joint failures.

arXiv Computer VisionIn-site articleCoupled or Independent? Fault Coupling in Agentic Medical Vision Systems

DyCraft: An Agentic System for Long-Horizon 4D Scene Generation

DyCraft breaks long-horizon animation generation into planning, static scene construction, and 3D storyboard keyframes, enabling coding agents to build and debug local motion step by step and reducing physical violations such as interpenetration and lack of support.

arXiv Computer VisionIn-site articleDyCraft: An Agentic System for Long-Horizon 4D Scene Generation

Mage-VL-Cowork: A Full Training Recipe to Learn Hybrid Computer Use From Evolved Harnesses

Mage-VL-Cowork is a 4B hybrid computer-use agent trained via a five-stage progressive curriculum, achieving 66.0% on OSWorld-Verified—state-of-the-art at its parameter size. Its data generation framework, Mage-CUA-Forge, synthesizes 293,806 tasks across 6,915 skills while cutting per-task cost by 82.6%.

arXiv Computational LinguisticsIn-site articleMage-VL-Cowork: A Full Training Recipe to Learn Hybrid Computer Use From Evolved Harnesses
Policy

Introducing the Grid Protectors Program

OpenAI launches a national program to work with utilities so future compute campuses can reduce demand during grid strain, supporting reliability and protecting consumer electricity costs.

OpenAI NewsIn-site articleIntroducing the Grid Protectors Program

Veloxity: Towards Memory-Safe UAV Embedded Firmware Using the Rust Programming Language

Veloxity is an open-source UAV autopilot written in Rust, aiming to advance memory-safe languages in aerospace. It demonstrates functional parity and improved performance compared to C++ autopilots in simulation, hardware testbenches, and outdoor flight tests.

arXiv RoboticsIn-site articleVeloxity: Towards Memory-Safe UAV Embedded Firmware Using the Rust Programming Language

Confidence-Modulated Safety Filtering for End-to-End Policies

The paper proposes a filter-agnostic pipeline that uses the policy's own confidence to set a safety filter's conservativeness at runtime: inside the filter's operational design domain (ODD), where every setting is safe, confidence buys performance; beyond it, low confidence adds margin. Preliminary F1TENTH simulations show the modulated filter combining the strengths of permissive and conservative filters.

arXiv RoboticsIn-site articleConfidence-Modulated Safety Filtering for End-to-End Policies

OpenAI appoints Dan Driscoll as Senior Counselor for National Security Modernization

OpenAI has appointed former Secretary of the Army Dan Driscoll to the newly created role of Senior Counselor for National Security Modernization, advising on responsible AI for national and cyber security and working with U.S. allies to advance democratic AI.

OpenAI NewsIn-site articleOpenAI appoints Dan Driscoll as Senior Counselor for National Security Modernization
Research

The First Few Tokens Matter: Understanding Chain-of-Thought Controllability

A new paper links models' limited control over their chain-of-thought to an entropy collapse in the first few tokens after the opening thinking tag. Inference-time interventions on these initial tokens—prefilling or banning frequent opening tokens—improve average compliance by 34-41 percentage points and mainly recover addition and style controllability, while suppression controllability responds differently.

arXiv AIIn-site articleThe First Few Tokens Matter: Understanding Chain-of-Thought Controllability

DiRAFT: Privacy-Aware Financial Transcript Synthesis from Aggregate Discourse Priors

The paper introduces DiRAFT, a method for synthesizing financial transcripts from aggregate concept-transition priors, and evaluates discourse fidelity and residual confidentiality risks. Human review of 1,400 investor-call outputs shows stochastic sampling yields fewer matches to non-public facts than greedy and top-3 sampling, though no formal privacy guarantee is established.

arXiv AIIn-site articleDiRAFT: Privacy-Aware Financial Transcript Synthesis from Aggregate Discourse Priors

Contact Imitation Interface: Transferring Dexterous Skills by Showing How Contact Unfolds

The paper presents Contact Imitation Interface (CIMI), a framework with contact-grounded demonstration and learning interfaces that transfers robot-free human dexterous demonstrations to robot hands, improving mean final-stage success by at least 50 percentage points over baselines across six hand-task pairs on LEAP and Wuji.

arXiv RoboticsIn-site articleContact Imitation Interface: Transferring Dexterous Skills by Showing How Contact Unfolds

From Teleoperation to Long-Horizon Dexterity: Evaluating Visuomotor Skills and Offline Proxies for Dexterous Manipulation

Accepted at the CoRL 2026 Workshop, this paper evaluates 281 imitation-learning policies across roughly 1,700 real-robot trials and finds that observation and action representations matter far more than optimizer or sampling choices, while most offline metrics fail to predict real-world success.

arXiv RoboticsIn-site articleFrom Teleoperation to Long-Horizon Dexterity: Evaluating Visuomotor Skills and Offline Proxies for Dexterous Manipulation

WorldBridge: Bridging Video and Action for Task Generalization in World Action Models

Researchers introduce WorldBridge, a World Action Model with separate video and action experts linked by structured attention and trained with joint flow matching. Providing action-free videos of new tasks lifts out-of-distribution success on RoboTwin 2.0 from 25.07% to 55.90%, and one labeled demonstration per task raises it to 70.70%.

arXiv RoboticsIn-site articleWorldBridge: Bridging Video and Action for Task Generalization in World Action Models

Let LiDAR Decide Where to Look: Geometry-Measured Token Selection for Efficient Sparse Latent World Models

The paper introduces GATS, which uses LiDAR and radar point projections to select a small fraction of image tokens before the vision backbone runs, cutting end-to-end FLOPs by up to 18x while outperforming the dense baseline on nuScenes and NAVSIM and substantially reducing collision rates.

arXiv Computer VisionIn-site articleLet LiDAR Decide Where to Look: Geometry-Measured Token Selection for Efficient Sparse Latent World Models

AquaProbe: Efficient Adaptation of Frozen Vision Models for Underwater Images

AquaProbe measures how the domain gap shifts with water type and imaging range, then allocates a fixed LoRA budget proportionally, improving segmentation by 1.9 mIoU over uniform allocation at the same 0.59M parameter budget and reaching 70.6 mIoU on the SUIM split.

arXiv Computer VisionIn-site articleAquaProbe: Efficient Adaptation of Frozen Vision Models for Underwater Images

EgoHSI3R: Online 4D Hand-Scene Interaction Reconstruction from Egocentric Video

EgoHSI3R is a unified feed-forward framework that reconstructs 4D hand-scene interactions online from egocentric monocular video, jointly estimating camera motion, dense scene geometry, world-space hands, and per-vertex contact. It achieves the lowest world-space hand joint error on ARCTIC among evaluated methods while offering faster inference than existing world-space hand reconstruction pipelines.

arXiv Computer VisionIn-site articleEgoHSI3R: Online 4D Hand-Scene Interaction Reconstruction from Egocentric Video

Better Retrievers, Worse Labels: How Incomplete Relevance Judgments Distort Abstention Evaluation in Conversational RAG

The paper shows that using the presence of gold documents as a proxy for answerability in conversational RAG benchmarks distorts abstention evaluation: stronger retrievers retrieve more contexts without gold documents that nonetheless contain the answer, dropping the shortcut's precision from .68 to .10. The authors recommend scoring refusals against content labels rather than gold presence.

arXiv Computational LinguisticsIn-site articleBetter Retrievers, Worse Labels: How Incomplete Relevance Judgments Distort Abstention Evaluation in Conversational RAG

Adaptive Personalization Under Preference Epistemic Uncertainty

The paper formalizes adaptive personalization under preference epistemic uncertainty and introduces a multilingual benchmark with 250 personas, 5 languages, and 25K conversational turns. Across eight models, it finds a recognition-generation gap: models select appropriate strategies when choosing among options but suffer a 35-46% drop in strategy calibration during open-ended generation, systematically defaulting to exploit-first behavior. Targeted training narrows the user acceptability generalization gap by about one third.

arXiv Computational LinguisticsIn-site articleAdaptive Personalization Under Preference Epistemic Uncertainty

Qistas: A Pre-Registered, Portable Protocol for Fairness Evaluation of Large Language Models

Qistas is a pre-registered, portable fairness evaluation protocol instantiated first for Iraq. It measures only procedural symmetry, differential refusal, stereotype reliance, and terminological framing, using paired designs and rigorous statistics. A semantic review of a 58-item seed bank found 24 of 25 template groups failed at least one check, exposing common hazards in counterfactual fairness benchmarks.

arXiv Computational LinguisticsIn-site articleQistas: A Pre-Registered, Portable Protocol for Fairness Evaluation of Large Language Models

TensorSSM: Memorizing Spatial Position in Dynamic State Space for Tensor Core Generation

TensorSSM introduces an implicit neural representation for functional tensor decomposition that encodes mode-wise coordinates as coarse-to-fine spatial tokens via a K-ary rule and extends a state space model to capture cross-mode spatial dependencies, improving reconstruction accuracy and storage efficiency without strict spatial assumptions.

arXiv Machine LearningIn-site articleTensorSSM: Memorizing Spatial Position in Dynamic State Space for Tensor Core Generation

Near-Optimal First-Order Regret for Bandits with Feedback Graphs

This paper studies first-order regret guarantees for adversarial bandits with a fixed known directed feedback graph. For strongly observable graphs, the authors bridge the gap between the minimax regret Θ̃(√(αT)) and the best existing first-order bound Õ(√((κ+1)L⋆)) by achieving an expected regret of Õ(√(αL⋆)+α), where α is the independence number and L⋆ is the cumulative loss of the best arm.

arXiv Machine LearningIn-site articleNear-Optimal First-Order Regret for Bandits with Feedback Graphs

Certified Behavioral Retention for Continual Adaptation with Held-Out Anchors

The paper introduces Certified Anchor-Based Adaptation (CABA), a candidate-agnostic post-update gate that uses held-out historical anchors on frozen representations to keep policy behavior changes within a declared budget, demonstrating a retention-plasticity trade-off in a UCI HAR proxy study.

arXiv Machine LearningIn-site articleCertified Behavioral Retention for Continual Adaptation with Held-Out Anchors

Unsupervised Feature Selection Using Local Intrinsic Dimensionality -- Extended Version

This paper proposes LIDFS, an unsupervised feature selection method based on Local Intrinsic Dimensionality (LID). The method transposes the data matrix so that each feature is treated as a point, and estimates the LID of each feature by maximum likelihood from distances to its k nearest neighboring features. Features with lower LID exhibit greater local discriminability and receive higher importance scores. A refinement step retains one representative per feature neighborhood and demotes locally redundant features. The result is a complete ranking from which any number of features can be selected without rerunning the procedure. LIDFS does not use label or class-count information and runs in O(d²n) time. On 23 high-dimensional datasets, LIDFS ranks first in supervised evaluations and fir…

arXiv Machine LearningIn-site articleUnsupervised Feature Selection Using Local Intrinsic Dimensionality -- Extended Version

Bayesian Distribution System State Estimation with Flow Matching

PoSE keeps the Bayesian structure of distribution system state estimation and uses a flow-matching neural network only to learn the prior distribution of substation injections, while power flow, topology and measurements enter through a noise-aware Kalman-type correction at sampling time. It improves CRPS by 6.7-7.5% on the IEEE 33-bus feeder, yields well-calibrated uncertainty, and adapts to reconfigured topologies without retraining.

arXiv Machine LearningIn-site articleBayesian Distribution System State Estimation with Flow Matching
Models

Incongruence-Aware Evaluation of Multimodal Speech Emotion Recognition in Audio Large Language Models

The paper introduces a threshold-free incongruence score measuring disagreement between the textual and acoustic emotion channels in speech, and tests whether that disagreement degrades emotion recognition in audio-capable large language models across three corpora.

arXiv AIIn-site articleIncongruence-Aware Evaluation of Multimodal Speech Emotion Recognition in Audio Large Language Models

Circumventing the Computational Hardness of Computational Arbitrage

The paper formalizes computational arbitrage in AI model marketplaces as an allocation problem and proves it is generally NP-hard, but shows the optimal allocation can be computed efficiently under a natural diminishing-returns condition on model performance versus inference spending. It also introduces a sample-efficient method for estimating model performance and derives several economic insights.

arXiv AIIn-site articleCircumventing the Computational Hardness of Computational Arbitrage

STAR: Steering Action Representations during Inference

Researchers propose STAR, a training-free inference method for vision-language-action models that applies an exponential moving average over late-layer residual updates and adapts mixing strength using recent proprioceptive states, improving average task success by 4.0 percentage points on robot manipulation benchmarks.

arXiv RoboticsIn-site articleSTAR: Steering Action Representations during Inference

Sparse Autoencoders for Expert-Assisted Interpretation of Histopathology Foundation Model Embeddings

This paper investigates whether sparse autoencoders (SAEs) can decompose histopathology foundation model embeddings into sparse latent features suitable for expert-assisted interpretation, training vanilla and Batch-TopK SAE variants on NCT-CRC embeddings with CONCH as the primary model and UNI2-h as a robustness check. The learned features preserve most downstream tissue-class information and retrieve more coherent, label-concentrated patch groups than raw embedding dimensions or PCA components, though the authors stress that latent features are candidate concepts requiring pathological assessment rather than validated histopathological concepts.

arXiv Computer VisionIn-site articleSparse Autoencoders for Expert-Assisted Interpretation of Histopathology Foundation Model Embeddings

PathoJev: Efficient Pathology Decisions from Shared Visual Context

PathoJev is a pathology AI model that directly predicts answer probabilities for multiple structured questions about a slide by encoding it once and reusing recurrent states, accelerating whole-slide and region-level assessment. On three whole-slide benchmarks, the 9B and 27B models achieve mean accuracies of 77.2% and 78.1%, exceeding SlideChat by 5.4 and 6.3 percentage points. On WSI-Bench, the 9B model runs 9.1x faster than SlideChat, and shared context accelerates 64-question workloads by up to 11.2x.

arXiv Computer VisionIn-site articlePathoJev: Efficient Pathology Decisions from Shared Visual Context

Language Models Show Conformity Bias in RAG: When the Majority Endorses a Poisoned Claim

A study using 200 SciFact questions finds that Gemma-3-12b's ability to detect poisoned claims in RAG drops from 25% to 10% when six mutually agreeing poisoned documents dominate the retrieved evidence, suggesting a conformity-like bias.

arXiv Computational LinguisticsIn-site articleLanguage Models Show Conformity Bias in RAG: When the Majority Endorses a Poisoned Claim

CoFiBreak: A Coarse-to-Fine Diversity Reward for Effective Multi-Turn Jailbreak

CoFiBreak introduces a coarse-to-fine diversity reward for GRPO-trained automated red-team attackers, improving attack strategy diversity while preserving high attack success and low inference cost, with transfer to unseen frontier models.

arXiv Computational LinguisticsIn-site articleCoFiBreak: A Coarse-to-Fine Diversity Reward for Effective Multi-Turn Jailbreak

The Generative Gap Between Speculative Drafter and Continuous Diffusion Language Model

A new arXiv paper compares continuous diffusion language models (CDLMs) and speculative drafter architectures under identical training budgets without verification. Using LangFlow and adapted DSpark variants, it finds LangFlow outperforms on both GenPPL and MAUVE at sequence lengths 128 and 1024, while DSpark-RNN leads at length 16. Increasing denoising steps further improves LangFlow and can change the length-128 ordering.

arXiv Computational LinguisticsIn-site articleThe Generative Gap Between Speculative Drafter and Continuous Diffusion Language Model

Li²SAO: Lifting Light Single-Rollout Asynchronous RL Optimization via Group-Relative Actor-Critic Bootstrapping

Li²SAO introduces a lightweight asynchronous RL method that bootstraps a cold critic with group-relative advantages and returns, then switches to single-rollout GAE, matching or surpassing GRPO early and raising later performance without a pretrained value model.

arXiv Computational LinguisticsIn-site articleLi²SAO: Lifting Light Single-Rollout Asynchronous RL Optimization via Group-Relative Actor-Critic Bootstrapping

FARO: Function-Space Adaptive Return-to-Risk Optimization via Cross-Block Equal-Risk-Contribution Budget Allocation

FARO introduces a function-space adaptive optimizer that treats parameter partitioning into blocks as the decision unit, assigns geometry by realized counterfactual return, and reallocates per-block budgets using equal risk contribution (ERC) from portfolio theory. On a GPT-style model, FARO-Adaptive matches the best hand-designed recipe, and FARO-ERC stably lowers clean-data loss across 6 seeds by 0.0136 nats on average without added time or memory.

arXiv Machine LearningIn-site articleFARO: Function-Space Adaptive Return-to-Risk Optimization via Cross-Block Equal-Risk-Contribution Budget Allocation

Frozen in Time: How a Physics Foundation Model Forgets Under Sim-to-Real Adaptation

This study reveals catastrophic forgetting in Walrus-1.3B during a two-stage sim-to-real adaptation: finetuning on simulated cylinder flow for just 20 epochs increases one-step errors up to 35x, and subsequent real-data adaptation further erodes simulation performance. Replaying only 1% simulated data recovers most lost simulation performance while retaining effective adaptation.

arXiv Machine LearningIn-site articleFrozen in Time: How a Physics Foundation Model Forgets Under Sim-to-Real Adaptation

Don't Just Impute, Detect: Task-Aware Diffusion Pipelines for Edge Anomaly Detection

The paper identifies an "imputation bias" in the standard impute-then-detect pipeline for IIoT edge data loss and proposes the Task-Aware Diffusion Refiner (TADR), which refines imputations against a frozen downstream detector's loss, raising F1 by 0.338 on SMD and characterizing when the method helps and how fast it runs on edge hardware.

arXiv Machine LearningIn-site articleDon't Just Impute, Detect: Task-Aware Diffusion Pipelines for Edge Anomaly Detection

Automatic Construction of Strong Simple Baselines

A new study uses large language models to automatically build simple baselines for scientific deep learning tasks, recovering a median of 95.5% of the performance gap across 36 tasks from top journals and matching or exceeding the original model in 12 studies, offering a scalable check on deep learning contributions.

arXiv Machine LearningIn-site articleAutomatic Construction of Strong Simple Baselines

Quoting Kyle Kingsbury

In Jepsen's analysis of ParticleDB 1.1.4, Kyle Kingsbury warns that LLMs erode trust in documentation, software, and the organizations behind them, and asks how maintainers can prove quality in this new era.

Simon Willison's WeblogIn-site articleQuoting Kyle Kingsbury
Robotics

A Capability-Aware Backup-Recoverability Filter for Heterogeneous Multi-Robot Navigation

The paper introduces CA-BRF, a filter that constructs a capability-dependent backup-recoverable set in shared Cartesian space and coordinates admissible interpolations between candidate and backup actions, preserving pairwise safety and guaranteeing recursive feasibility and safety even when nominal safety-constrained optimization becomes infeasible; validated in simulations with up to 30 heterogeneous robots and experiments across five kinematic classes.

arXiv RoboticsIn-site articleA Capability-Aware Backup-Recoverability Filter for Heterogeneous Multi-Robot Navigation

Mechanistic Interpretability for End-to-End Self-Driving

Researchers apply mechanistic interpretability to end-to-end autonomous driving by training sparse autoencoders on a single-camera planner's activations, revealing interpretable features for route intent, speed, and braking that causally affect outputs and improve trajectory selection.

arXiv RoboticsIn-site articleMechanistic Interpretability for End-to-End Self-Driving

FlapSim: A Composable MuJoCo Simulation Platform and Reduced-Order Aerodynamic Model Hierarchy for Flapping-Wing Robotics

Researchers introduce FlapSim, an open-source modular MuJoCo platform for flapping-wing robotics, and benchmark five reduced-order aerodynamic configurations across 48 held-out scenarios. Careful timing alignment and harmonic reconstruction lift mean wrench fidelity from 0.6593 to roughly 0.966 while retaining about 65-68x CPU speedup over a quasi-steady blade-element reference.

arXiv RoboticsIn-site articleFlapSim: A Composable MuJoCo Simulation Platform and Reduced-Order Aerodynamic Model Hierarchy for Flapping-Wing Robotics

AutoHebb: Self-Modifying Neural Networks that Zero-Shot Infer and Continually Learn

AutoHebb is a class of online-learning feedforward networks that combine meta-learning, Hebbian plasticity, and autonomous learning control to achieve zero-shot inference and continual learning, outperforming modern models and enabling reverse-engineering.

arXiv Machine LearningIn-site articleAutoHebb: Self-Modifying Neural Networks that Zero-Shot Infer and Continually Learn
Chips

UFMF: Radar-Guided Kinematic Priors for Real-Time 4D Occupancy Forecasting

UFMF is a real-time 4D occupancy forecaster that achieves 122 FPS on a single RTX 4090 and 0.45 fast-displacement mIoU on nuScenes by using a radar-guided velocity-aware dynamic prior and auxiliary training constraints, outperforming occupancy world models at a fraction of the cost.

arXiv Computer VisionIn-site articleUFMF: Radar-Guided Kinematic Priors for Real-Time 4D Occupancy Forecasting