Research

• Computer Science > Artificial Intelligence [Submitted on 17 Feb 2026] Title:Quantifying construct validity in large language model evaluations View PDF HTML (experimental)Abstrac

• Computer Science > Artificial Intelligence [Submitted on 16 Feb 2026] Title:ResearchGym: Evaluating Language Model Agents on Real-World AI Research View PDF HTML (experimental)Ab

• Computer Science > Artificial Intelligence [Submitted on 16 Feb 2026] Title:Secure and Energy-Efficient Wireless Agentic AI Networks View PDF HTML (experimental)Abstract:In this

• Computer Science > Artificial Intelligence [Submitted on 17 Feb 2026] Title:When Remembering and Planning are Worth it: Navigating under Change View PDF HTML (experimental)Abstra

• Computer Science > Artificial Intelligence [Submitted on 17 Feb 2026] Title:World-Model-Augmented Web Agents with Action Correction View PDF HTML (experimental)Abstract:Web agent

• Study shows cannabis-infused drinks can cut alcohol consumption by nearly 50%.\n• Researchers from University at Buffalo highlight cannabis as a harm‑reduction tool for heavy dri

• Email Bluesky Facebook LinkedIn Reddit Whatsapp X Credit: Cheng Xin/Getty Chinese research organizations can no longer take part in most of the research grants funded by Horizon

• Nature article on EGFR blockade response in colorectal cancer corrected figure duplication. • Micrograph in Extended Data Fig. 8 mistakenly duplicated from Fig. 10, misrepresenti

• UKRI paused medical, biological, and physical science grants amid policy review, unsettling researchers. • Short‑term contract holders risk career setbacks due to funding freeze.

• Science Robotics, Volume 11, Issue 111, February 2026.

• Science Robotics, Volume 11, Issue 111, February 2026.

• Horses with ≥30 minutes REM sleep daily outperform peers in field learning tasks. • Short REM periods reduce perseverance and performance during demanding training. • New field‑a

• Legal threats historically silenced detailed, negative consumer reviews online in. • Removing these threats instantly spurred longer, more candid feedback from U.S. consumers. •

• A newly examined prehistoric shark from the age of dinosaurs provides surprising insights into the early evolution of modern sharks. • It cannot be confidently assigned to any sh

• Seal pups exhibit turn-taking in vocal exchanges, mirroring human conversational patterns. • Their calls converge over time, becoming more similar when pups interact closely. • R

• Share Link copied to clipboard! • Content types Industry trends Topics AI and agents Defending against advanced tactics Security management Security operations SIEM and XDR Secur

• New research claims wormholes are temporal mirrors, not interstellar tunnels The ‘mirror’ framework addresses the black hole information paradox, a conflict between quantum mecha

• The post Science on the Double: How an AI-Powered ‘Digital Twin’ Accelerates Chemistry and Materials Discoveries appeared first on Berkeley Lab News Center . • The post What Is a

• What Is a Digital Twin? • Video AI Digital twins are transforming how scientists study and improve complex systems - reducing the time between discovery and delivery. • Learn wha

• Mallory LindahlTuesday, February 17, 2026Print this page. • Three faculty members in Carnegie Mellon University’s School of Computer Science are among five CMU researchers select

• Microplastics detected in droppings of freshwater birds across Europe • Study led by University of Glasgow, published in Environmental Research • Findings confirm widespread pres

• James Cook University (JCU) research argues Australians urgently need better education about heat to prepare for longer, hotter and more dangerous heat waves driven by climate ch

• Tropical forests produce 2.4 million liters of rainfall per hectare annually. • Study quantifies rainfall generation, equating to Olympic-sized pool per hectare. • Findings highl

• Researchers at the Department of Energy’s Oak Ridge National Laboratory are helping to pave a path for the eventual discovery of dark matter. • With new approaches to measurement

• Subjects Health policy Medical humanities Less than 2% of artificial intelligence devices authorized by the US Food and Drug Agency are prognostic, with prediction horizons rangi

• Home Systems & Design Low Power - High Performance Manufacturing, Packaging & Materials Test, Measurement & Analytics Auto, Security & Enabling Technologies Special Reports Busin

• Computer Science > Artificial Intelligence [Submitted on 26 Jan 2026] Title:A Geometric Taxonomy of Hallucinations in LLMs View PDF HTML (experimental)Abstract:The term ‘hallucin

• Computer Science > Artificial Intelligence [Submitted on 21 Jan 2026] Title:Agentic AI for Commercial Insurance Underwriting with Adversarial Self-Critique View PDF HTML (experim

• Computer Science > Artificial Intelligence [Submitted on 30 Jan 2026] Title:AST-PAC: AST-guided Membership Inference for Code View PDF HTML (experimental)Abstract:Code Large Lang

• Computer Science > Artificial Intelligence [Submitted on 22 Jan 2026] Title:BotzoneBench: Scalable LLM Evaluation via Graded AI Anchors View PDF HTML (experimental)Abstract:Large

• Computer Science > Artificial Intelligence [Submitted on 2 Feb 2026] Title:DPBench: Large Language Models Struggle with Simultaneous Coordination View PDF HTML (experimental)Abst

• Computer Science > Artificial Intelligence [Submitted on 3 Feb 2026] Title:General learned delegation by clones View PDF HTML (experimental)Abstract:Frontier language models impr

• Computer Science > Artificial Intelligence [Submitted on 4 Feb 2026] Title:Human-Centered Explainable AI for Security Enhancement: A Deep Intrusion Detection Framework View PDF H

• Computer Science > Artificial Intelligence [Submitted on 28 Jan 2026] Title:Intelligence as Trajectory-Dominant Pareto Optimization View PDF HTML (experimental)Abstract:Despite r

• Computer Science > Artificial Intelligence [Submitted on 29 Jan 2026] Title:Lang2Act: Fine-Grained Visual Reasoning through Self-Emergent Linguistic Toolchains View PDF HTML (exp

• Computer Science > Artificial Intelligence [Submitted on 3 Feb 2026] Title:MAPLE: A Sub-Agent Architecture for Memory, Learning, and Personalization in Agentic AI Systems View PD

• Computer Science > Artificial Intelligence [Submitted on 29 Jan 2026] Title:NL2LOGIC: AST-Guided Translation of Natural Language into First-Order Logic with Large Language Models

• Computer Science > Artificial Intelligence [Submitted on 29 Jan 2026] Title:PlotChain: Deterministic Checkpointed Evaluation of Multimodal LLMs on Engineering Plot Reading View P

• Computer Science > Artificial Intelligence [Submitted on 5 Feb 2026] Title:ProMoral-Bench: Evaluating Prompting Strategies for Moral Reasoning and Safety in LLMs View PDF HTML (e

• Computer Science > Artificial Intelligence [Submitted on 23 Jan 2026] Title:Scaling the Scaling Logic: Agentic Meta-Synthesis of Logic Reasoning View PDF HTML (experimental)Abstr

• Computer Science > Artificial Intelligence [Submitted on 29 Jan 2026] Title:Stay in Character, Stay Safe: Dual-Cycle Adversarial Self-Evolution for Safety Role-Playing Agents Vie

• Computer Science > Artificial Intelligence [Submitted on 5 Feb 2026] Title:TemporalBench: A Benchmark for Evaluating LLM-Based Agents on Contextual and Event-Informed Time Series

• TemporalBench offers a multi-domain benchmark for temporal reasoning in LLM agents. • Four-tier taxonomy tests historical structure, context-free, contextual, and event-condition

• Computer Science > Artificial Intelligence [Submitted on 27 Jan 2026] Title:Variation is the Key: A Variation-Based Framework for LLM-Generated Text Detection View PDF HTML (expe

• Computer Science > Artificial Intelligence [Submitted on 2 Feb 2026] Title:X-Blocks: Linguistic Building Blocks of Natural Language Explanations for Automated Vehicles View PDF H

• Some continue to criticize Prusa Research’s new Open Community License, but why? • The backstory: Last December, Prusa Research announced anew license for their products, specifi

• The post Basics2Breakthroughs: Studying the Hydraulic Mechanism in Jumping Spiders appeared first on Berkeley Lab News Center .

• GABRIEL is a new open-source toolkit from OpenAI that uses GPT to turn qualitative text and images into quantitative data, helping social scientists analyze research at scale.

• Gemini 3 Deep Think: Advancing science, research and engineering Feb 12, 2026 Our most specialized reasoning mode is now updated to solve modern science, research and engineering

• The post Crunching Big Data Into 3D Images Accelerates Discovery appeared first on Berkeley Lab News Center .

• Safety review completed at South African research reactor The five-day six-person Safety Review Mission on Ageing Management and Continued Safe Operation was at the invitation of

• The post A ‘Robot Pizza Chef’ Serving Up Better Quantum Computers appeared first on Berkeley Lab News Center .

• The post New AI Sensor ‘Sniffs’ Out Spectral Targets appeared first on Berkeley Lab News Center .

• Carnegie Mellon’s Cagan, Jahanian, and alumnus Pitel elected to the National Academy of Engineering 2026 class. • Cagan leads Mechanical Engineering, pioneering design automation

• News Views Podcast Learn team about contribute republish AIhub resources AIhub events News Views Podcast Learn News Views Podcast Learn Sven Koenig wins the 2026 ACM/SIGAI Autono

• US funding for research into recycling used nuclear fuel The Department of Energy (DOE) noted that less than 5% of the potential energy in the USA’s nuclear fuel is extracted aft

• Space is dominated by plasma, the most common state of matter in the cosmos. • Modeling plasma is tough: fluid dynamics, turbulence, and Maxwell’s equations all intertwine. • The

• The Accessible Platform Architectures Working Group has published the first draft of the Group Note titled Cognitive Accessibility Research Modules . • This set of modules looks

• The post Berkeley Lab Leads Effort to Build AI Assistant for Energy Materials Discovery appeared first on Berkeley Lab News Center .

• Open source is now the dominant strategy for Chinese AI firms, driving large‑scale deployment and integration. • DeepSeek leads Hugging Face followers, while Qwen ranks fourth, s