Skip to content
AI NEWS JOURNAL
AI, IN PERSPECTIVE

Search AI news

Search headlines, summaries and articles to connect the dots.

Newest first

Which Successes Make Good References? Action-Sequence Alignment for Multi-Turn On-Policy Distillation

arXiv:2611.13135v1 Announce Type: new Abstract: On-policy distillation (OPD) reduces train-inference distribution mismatch by supervising a student on its own trajectories. For multi-turn tool-use agents, conditioning the teacher on a validated same-task successful trajectory as a reference can provide useful hindsight, but different successes may reach incompatible interaction states and offer misleading supervision. We cast successful-reference selection as a free-target optimal transport problem, which reduces to assigning each failed trajectory its most compatible successful reference. Because latent state paths are hidden, we use action-sequence alignment as an observable proxy for compatibility and instantiate it with longest common subsequence (LCS), which identifies a shared ordered backbone of actions while allowing recoverable detours. This perspective makes trajectory compatibility the central criterion for reference selection in multi-turn OPD. Experiments on ALFWorld and the airline domain of $\tau^3$-bench show that LCS is the only evaluated reference selector to improve over reference-free OPD on both ALFWorld overall scores and all three airline metrics. Without further optimization, the airline-trained models also transfer to the retail domain, where LCS improves all three metrics.

b12229

A fix was implemented to ensure the OpenAI-compatible chat completions endpoint correctly processes JSON schema when provided at the top level in response_format.schema, rather than requiring it to be nested within response_format.json_schema.schema. The update also improves error messages, documents the correct usage in the README, and adds test cases to verify behavior, resolving issue #10732.

Clinically Informed Colour Encoding Improves the Localisation and Stability of Explanations in Deep Learning ECG Models: External Validation in a UK Biobank Epilepsy Cohort

arXiv:2611.09971v1 Announce Type: new Abstract: Objective: It is not known whether the input representation governs the quality of explanations of deep learning models for electrocardiogram analysis. Risk of dangerous ventricular repolarisation is judged from the QT-interval and the shape of the T-wave, features that are hard to assess visually and harder still to verify in a model's reasoning. We evaluate a pseudo-colour encoding filling the area beneath the QT interval, its colour sequence designed from clinical electrophysiology knowledge and mapped to a validated risk prediction model, the QT-nomogram. Methods: Two identically configured residual networks were trained on single-heartbeat images from a clinical trial, differing only in whether pseudo-colour encoding was applied, so that any difference is attributable to the chromatic channel alone. To externally validate the effect, both models were applied without fine-tuning or threshold tuning to a previously unseen epilepsy cohort from the UK-Biobank at two visits, and eight explanation methods compared on faithfulness, localisation and stability. Results: The pseudo-colour model achieved areas under the receiver operating characteristic curve of 0.903 and 0.914 at the first and follow-up visits, against 0.719 and 0.720 for the monochrome model, recovering three to five times as many at-risk heartbeats at comparable specificity. Pseudo-colour encoding improved explanation localisation and stability for almost every method, and the improvement reproduced across both visits. Conclusion: The input representation is a controllable and previously overlooked determinant of both predictive performance and explanation quality. Significance: Treating interpretability as a representation-design problem offers a route towards explanations clinicians can check against known electrophysiology.

CulturalCrosswords: A Multilingual Benchmark comprised of Answer Constraints and Grid Completion

arXiv:2611.10044v1 Announce Type: new Abstract: When several answers fit a riddle, exact-match scoring can penalise a valid response. CulturalCrosswords tests how answer constraints affect large language models (LLMs) solving cultural riddles, whose solutions draw on knowledge specific to a community. We constructed 53 crosswords from crowdsourced riddles in 42 languages and evaluated 11 LLM configurations on five versions of each clue set: plain batches, batches with lengths, batches with lengths and partial-answer hints, and crosswords requesting an answer list or a completed grid. On 962 clues with alternative answers, lengths improve agreement with the reference by 22.8 percentage points, compared with 8.4 points when accepted alternatives also receive credit. Part of the gain therefore appears to come from selecting a required answer form. Crossword input increases the share of all crossings with compatible answers by 9.7 points; requiring a completed grid reveals large differences between models. The benchmark distinguishes agreement with a reference from the ability to produce answers that fit together and occupy the correct cells.

FirstSwap: Pre-Consensus Trade Approvals for Decentralized Market Design

arXiv:2611.10254v1 Announce Type: new Abstract: Decentralized exchanges typically give traders no execution guarantee until a block is finalized. In the meantime, external prices move, traders strategically delay submission, and block producers can influence inclusion and ordering. We introduce FirstSwap, a DEX built on a constant-function market maker pricing curve that gives traders an early pre-approval before consensus, guaranteeing inclusion and execution within their submitted price limit, with final ordering determined by limit prices rather than by a block producer. We demonstrate the benefits of trade pre-approvals in an economic model combining auction theory with CEX-DEX arbitrage. When the CEX price changes gradually, FirstSwap rewards earlier traders with better prices, discouraging strategic delay and yielding higher social welfare and gains to AMM liquidity providers than a batch auction. When the CEX price jumps abruptly, FirstSwap yields higher welfare and smaller AMM losses than fair ordering. The central challenge is delivering fast trade pre-approvals without consensus. We introduce a decentralized, scaled admissibility protocol under which each validator reserves only a fraction of the available liquidity at every price. A trader obtains approval in a single round of messages with a quorum of validators. Somewhat surprisingly, despite Byzantine validators who may approve trades arbitrarily, we show that all quorum-certified trades are jointly admissible and execute within their limits. Our protocol extends naturally to a setting where buy and sell orders can be pre-approved against each other before a block is finalized.

Gemini 4 Carbon now available on AI Gateway

Gemini 4 Carbon, a model designed for complex software engineering, agentic work, and cybersecurity defense, is now available on Vercel's AI Gateway with a 50% discount through November 30, 2026. It features a 1M-token context window, enhanced capabilities in coding, full-stack development, scientific reasoning, and vulnerability detection, along with improved multi-step planning and collaboration under ambiguous requirements. The model can be accessed using the identifier 'google/gemini-4-carbon' in supported tools after setting up the Vercel CLI. AI Gateway offers unified API access, usage tracking, cost management, and performance optimizations, with no markup or platform fees, including for BYOK requests.

How CrossnoKaye grounds physical AI with Amazon Bedrock AgentCore

CrossnoKaye brought physical AI to heavy industry with Amazon Bedrock AgentCore. Its ATLAS platform and AURA agentic system ground AI reasoning in live plant data and run verifiable Python in isolated AgentCore sandboxes, giving plant engineers checkable, math-backed answers instead of model guesses.

How Many Demonstrations Does a VLA Need? On the Sample Efficiency of VLA Finetuning

arXiv:2611.09964v1 Announce Type: new Abstract: Pretrained vision language action (VLA) models show strong zero-shot performance, but are rarely deployed directly and are often first post-trained on demonstrations from the target setting. Yet VLAs are typically compared by zero-shot or full-data fine-tuned success, revealing little about the data and compute required for adaptation. We study adaptation efficiency for six VLAs ($\pi_{0}$, $\pi_{0.5}$, and GR00T N1-N1.7) on four LIBERO suites, varying both demonstrations and training steps. This reveals differences hidden by full-data fine-tuning: models within two points on the full LIBERO-Spatial dataset differ by up to $3\times$ in success with 50 demonstrations. $\pi_{0.5}$ needs $2$--$5\times$ fewer demonstrations than $\pi_{0}$ and is the most sample-efficient model at every budget, with the largest gap on the long-horizon suite. Within GR00T, N1.5 adapts more efficiently than N1.6 and N1.7 on three of four suites, despite their stronger zero-shot performance. We study one pretraining factor behind these differences by freezing the VLM during finetuning: this severely degrades the $\pi$ models, whose VLM is jointly trained with the action expert during pretraining, but barely affects GR00T models. Finally, validation loss on held-out demonstrations tracks closed-loop success within a suite, but understates how much more sample-efficient $\pi_{0.5}$ is than $\pi_{0}$.

How skills and agents work in AWS Quick

Repeatable tasks like status updates, release notes, and ticket creation can be encoded in AWS Quick through agents and skills. This post covers how agents, skills, and your data fit together in AWS Quick, how to build and share skills for your team, and how the official catalog and custom agents help you use them.

IBM builds on hybrid cloud and governance to advance its enterprise AI operating model

Enterprises are moving beyond AI pilots, putting greater emphasis on governance, production readiness and measurable business outcomes. Research from Deloitte highlights a significant gap between enterprise AI ambition and execution: 54% of organizations plan to move at least 40% of their AI initiatives into production within six months. Yet only 25% have done so today. IBM […] The post IBM builds on hybrid cloud and governance to advance its enterprise AI operating model appeared first on SiliconANGLE.

Indie Game Development with AI

Article URL: https://www.sandromaglione.com/articles/indie-game-development-with-ai Comments URL: https://news.ycombinator.com/item?id=50432523 Points: 22 # Comments: 16

Meet the Sad Wives of AI

Article URL: https://www.wired.com/story/meet-the-sad-wives-of-ai/ Comments URL: https://news.ycombinator.com/item?id=50438429 Points: 27 # Comments: 13

n8n@2.48.0

The n8n 2.48.0 release addresses multiple bugs, including fixes for AI builder, compression node, core functionality, editor improvements, and RabbitMQ node issues. New features include enhanced assistant models, Microsoft Work IQ OAuth2 credential support, app bindings UI, migration report enhancements, and improved user experience elements like chat message unification and credential modal restrictions.

n8n@3.2.0

The n8n 3.2.0 release introduces several bug fixes addressing issues in AI Builder, core functionality, editor UI, and node-specific behaviors, including improved handling of JSON, XML, CSV, and YAML responses, enhanced security through expanded IP blocking, and better error reporting. New features include support for Microsoft Work IQ OAuth2 credentials, a new App bindings UI, and updated default Assistant models. The release also reverts some previous changes due to compatibility or stability concerns.

NVIDIA Nemotron 3.5 Super VL is now available in Amazon SageMaker JumpStart

NVIDIA Nemotron 3.5 Super VL, a 120B-parameter multimodal vision-language model, is now available in Amazon SageMaker JumpStart. Deploy the NVFP4 or BF16 variant in a few steps, then run text, image, and video inference for document intelligence, computer-use agents, and multimodal agentic workloads.

Pay for Hugging Face Hub with your Google Cloud Account

Hugging Face has launched integration with the Google Cloud Marketplace, allowing organizations to link their Hugging Face accounts to Google Cloud for consolidated billing. This enables teams to manage Hugging Face usage—such as GPU workloads, model deployment, and inference services—through their existing Google Cloud billing account, eliminating the need for separate vendor onboarding or credit card payments. The integration supports all paid services, including Jobs, Inference Endpoints, and additional storage, with charges appearing on the Google Cloud invoice. Admins can subscribe via the Google Cloud Marketplace, link their organization, and await approval before usage is billed through Google Cloud. Once approved, the Google Cloud account becomes the active payment method, and usage can be monitored in both Hugging Face’s billing settings and Google Cloud’s billing reports.

Retrieval-Grounded Multi-View Reasoning for Clinically Reliable Radiology Report Generation

arXiv:2611.09993v1 Announce Type: new Abstract: Chest radiograph report generation is hampered by retrieved reports that resemble the current image but are clinically inconsistent with it, and by diagnostic-label imbalance. We present RG-MVR, a retrieval-grounded framework comprising Retrieval-Augmented Evidence Grounding, which combines dense memory summaries with visually gated sparse retrieved evidence; Adaptive Multi-view Evidential Reasoning, which weights image, memory, and retrieved evidence per case; and a Clinical Prompt Learner, which models finding mention separately from positive, negative, or uncertain status. On MIMIC-CXR, RG-MVR achieved a clinical-label F1 of 0.483 and a BLEU-4 of 0.118 (IU X-ray: 0.202 and 0.110). In an exploratory external cohort of 22 examinations, clinical-label agreement declined markedly (F1, 0.071) and unsupported statements remained common, although physicians found fewer unsupported temporal comparisons (9 vs. 20 cases) and device-related statements (7 vs. 15) than with the baseline. Evidence-aware retrieval and structured diagnostic prompting may reduce specific reporting errors, but broader external validation is required.