AI Research

Exploring the world's languages with Language Explorer

Natural Language Processing

Continuous Batching Beats Static and Dynamic by Up to 2.91× in vLLM

For anyone running an inference server: a zero-variance control, five traffic traces, and an engine-vs-scheduler split show where the gains come from. The post Continuous Batching Beats Static and Dynamic by Up to 2.91×…

Adobe Finds 68% of Creative AI Users Sometimes Settle for Results That Miss Their Vision

Adobe’s latest Creators’ Toolkit Report captures both the opportunity and the friction in creative AI: users report faster growth, but many still settle for results that miss their vision. Ahead of MAX, precision and…

Ask AI to Refine Your Ideas, Not Invent Them

Studies of short stories, metaphors, and 486 participants point one way, an 800-person experiment points the other, and the author weighs both. The post Ask AI to Refine Your Ideas, Not Invent Them appeared first on…

Liquid Energy Wants to Unfold an AI Data Center and Start Making Tokens

Liquid Energy’s NanoDC combines AI computing, folding solar panels, batteries, and Starlink in a factory-built system. The YC-backed startup wants to shorten the wait for inference capacity by bringing its own power.

How to Estimate the Recall of AI Agents with Capture-Recapture

A fisheries estimator from the 1890s, tested on 294 CUAD contracts, landed within 2 points of true recall for most agents with a 10% hand audit. The post How to Estimate the Recall of AI Agents with Capture-Recapture…

Quantile Regression Caught the House Price OLS Missed by $55K

On 88 house sales, OLS predicted $164K for a $219K home, while a 90% interval from 5th and 95th percentile models spanned $136K to $227K. The post Quantile Regression Caught the House Price OLS Missed by $55K appeared…

😺 Anthropic's AI cracks a puzzle

Adobe Says the Next AI Skill Isn’t Better Prompting. It’s Creative Direction.

Adobe says creative direction will become increasingly important as AI agents take on complex production tasks. Ahead of MAX 2026, the company is betting that Firefly, Creative Cloud, and third-party integrations can…

Why Jensen Huang's Affair Has Become an AI Chip Security Concern

Nvidia is investigating reports that CEO Jensen Huang was targeted by Chinese intelligence after he admitted to an affair. Here's what is known about the allegations and the company's response.

Anthropic's $3.4 Billion Power Deal Shows How Expensive AI Is Becoming

Anthropic reportedly agreed to spend $3.4 billion securing gas generators for its AI data centers. Here's why electricity shortages are becoming so expensive and what the purchase could mean for the environment.

Beyond Overlap: Hierarchical Graph Abstraction for Out-of-Core Vector Search

arXiv:2512.22838v3 Announce Type: replace-cross Abstract: At billion scale, approximate nearest neighbor search (ANNS) often becomes an out-of-core problem: the full vector collection and index structures exceed memory…

A Taxonomy-Driven Synthetic Dataset for Retrieval Effectiveness and Fairness Evaluation in Hiring

arXiv:2611.09178v1 Announce Type: cross Abstract: While machine learning hiring tools are ubiquitous, the need for privacy regulations and commercial sensitivity makes public benchmarking legally complex. We address…

PULSAR: Pooled Unified Late-Interaction Search and Retrieval for Enterprise Visual Document RAG

arXiv:2608.28572v2 Announce Type: replace Abstract: Institutional investors search visually dense pitch decks, board packs, and diligence materials that change hourly near deal closing. OCR followed by figure…

Structure and Lineage Evidence for Vietnamese Statutory Article Retrieval on DRiLL

arXiv:2611.09200v1 Announce Type: new Abstract: We study Vietnamese statutory Article retrieval on DRiLL using features for provision identity and local form, same-law support, and citation and lineage evidence. Across…

Generative Search and Recommendation at Shopify

arXiv:2611.08804v1 Announce Type: new Abstract: Search and recommendation at Shopify require retrieving relevant products from a changing catalog of billions of products under strict latency constraints. We present…

ENGRAM: From Agent Experience to Weights through Verified Consolidation

arXiv:2611.09472v1 Announce Type: cross Abstract: Language-model agents use experience through external retrieval and parameter learning. Consolidating stored experience into model weights requires deciding both which…

Unified Heterogeneous Feature Interaction via Effective-Rank Scaling for Recommender Systems

arXiv:2610.35318v4 Announce Type: replace Abstract: Large-scale industrial ranking models must jointly capture heterogeneous feature interactions, long-term user behaviors, and multi-scenario dependencies under…

SkinMind: Equipping Agentic RAG with Human-Inspired Graph Probing for Reliable Dermatology AI

arXiv:2611.08935v1 Announce Type: new Abstract: Reliable dermatology AI requires access to verifiable evidence distributed across heterogeneous textual and visual sources. However, conventional retrieval-augmented…

Knowledge Graph-based Retrieval-Augmented Generation for Schema Matching

arXiv:2501.08686v2 Announce Type: replace-cross Abstract: Traditional similarity-based schema matching methods fail to resolve semantic ambiguities and conflicts in complex, domain-specific mappings due to a lack of…