Artemis II Crew Trains on T-38
Artemis II astronauts trained on T-38 jets, which helps prepare them for the rigors of upcoming Moon missions and inspires the public by showing NASA's commitment to safe, successful human spaceflight.
report
Curious Now
Latest
What's new in science
Sales Research Agent and Sales Research Bench
Enterprises can now use an AI agent to quickly find sales data insights, rather than manually searching through CRM systems. This makes sales teams more efficient and helps leaders make better-informed decisions.
preprint
SubQuad: Near-Quadratic-Free Structure Inference with Distribution-Balanced Objectives in Adaptive Receptor framework
A new algorithm can efficiently analyze immune system data to identify rare but important immune cell types, helping doctors understand immune responses and diseases.
preprint
FRSICL: LLM-Enabled In-Context Learning Flight Resource Allocation for Fresh Data Collection in UAV-Assisted Wildfire Monitoring
Researchers developed a system to improve data collection by UAVs monitoring wildfires, which could help detect and respond to wildfires faster, reducing environmental damage.
preprint
Text Before Vision: Staged Knowledge Injection Matters for Agentic RLVR in Ultra-High-Resolution Remote Sensing Understanding
A new AI system can better understand complex satellite images by first learning relevant information through text, before analyzing the visual data. This could improve how we use satellite imagery to study the environment and plan infrastructure.
preprint
From Diagnosis to Inoculation: Building Cognitive Resistance to AI Disempowerment
Researchers found that AI assistants can subtly distort users' reality, values, and actions in harmful ways. This discovery highlights the importance of developing psychological resilience to potential AI manipulation.
preprint
From Subtle to Significant: Prompt-Driven Self-Improving Optimization in Test-Time Graph OOD Detection
Researchers developed a way to detect when a graph model is being used on data it wasn't trained on, helping ensure the reliability of graph AI systems in real-world applications.
preprint
An effective Genetic Programming Hyper-Heuristic for Uncertain Agile Satellite Scheduling
Researchers developed an advanced algorithm that helps schedule satellite operations even when there is uncertainty about factors like profit, resources, and weather. This could lead to more reliable and flexible Earth observation from satellites.
preprint
Pushing the Frontier of Black-Box LVLM Attacks via Fine-Grained Detail Targeting
Researchers found new ways to attack black-box AI language models by exploiting fine-grained details, which could lead to better security measures to protect these powerful systems from abuse.
preprint
Just KIDDIN: Knowledge Infusion and Distillation for Detection of INdecent Memes
Researchers developed a framework to better detect toxic content online by combining text and images, which could help make social media platforms safer for users.
preprint
Conv-FinRe: A Conversational and Longitudinal Benchmark for Utility-Grounded Financial Recommendation
Researchers developed a benchmark to evaluate AI systems that give financial advice, focusing on long-term utility rather than just imitating user behavior, which can be short-sighted in volatile markets.
preprint
Position-Aware Scene-Appearance Disentanglement for Bidirectional Photoacoustic Microscopy Registration
Researchers developed a technique to align microscope images taken from opposite scan directions, which could improve the speed and accuracy of photoacoustic imaging used for medical diagnosis and research.
preprint
Eigenmood Space: Uncertainty-Aware Spectral Graph Analysis of Psychological Patterns in Classical Persian Poetry
Researchers developed a new algorithm that can analyze patterns of emotion and psychology hidden within classical Persian poetry, offering a new way to understand the emotional lives of people long ago.
preprint
CLEF HIPE-2026: Evaluating Accurate and Efficient Person-Place Relation Extraction from Multilingual Historical Texts
A new CLEF evaluation lab will test how well AI systems can extract information about people and places from old texts written in different languages. This could help historians and journalists better understand the connections between historical figures and locations.
preprint
Beyond Needle(s) in the Embodied Haystack: Environment, Architecture, and Training Considerations for Long Context Reasoning
Researchers developed a new framework called $\infty$-THOR that can better understand and reason about long-term contexts in embodied AI systems, which could lead to more capable and contextually-aware AI assistants.
preprint
PETS: A Principled Framework Towards Optimal Trajectory Allocation for Efficient Test-Time Self-Consistency
A new framework called PETS helps AI models perform more consistently during testing, which could lead to more reliable and effective AI systems in real-world applications.
preprint
Conditional Flow Matching for Continuous Anomaly Detection in Autonomous Driving on a Manifold-Aware Spectral Space
Researchers developed a new method to help self-driving cars continuously detect and avoid rare, risky situations, improving safety for autonomous vehicles.
preprint
TAPO-Structured Description Logic for Information Behavior: Procedural and Oracle-Based Extensions
Researchers developed a new logical framework to model how people interact with information in a structured, dynamic way, which could improve information retrieval and recommendation systems.
preprint
Forecasting Anomaly Precursors via Uncertainty-Aware Time-Series Ensembles
Researchers developed a way to spot potential problems before they happen by analyzing patterns in data over time. This could help prevent system failures in fields like industry, finance, and cybersecurity by giving early warnings.
preprint
The Cascade Equivalence Hypothesis: When Do Speech LLMs Behave Like ASR$\rightarrow$LLM Pipelines?
Current speech language models can perform as well as combining speech recognition with language models, without explicit speech recognition. This means these speech models can produce accurate text output from audio inputs more efficiently.
preprint