The Trust Deficit: How AI Video Alters Perceptions
New research reveals that communicating via AI-mediated video, especially with realistic avatars, diminishes feelings of trust and confidence in judgments, even if it doesn’t improve lie detection.
Science
Seeing Beauty: An AI That Explains What Makes an Image Appealing
Researchers have developed a new framework for image aesthetic assessment that moves beyond simply scoring pictures to actually explaining why they are considered beautiful.
Seeing Through the Noise: A Smarter VINS for Challenging Environments
A new approach combines deep learning-powered optical flow with visual-inertial odometry to significantly improve state estimation in difficult conditions.
Web to Database: Automating Knowledge Capture from the Open Web
![The system navigates the complexities of information retrieval by deeply investigating specialized online resources-a process symbolized by [latex]\mathcal{C}0[/latex]-and reinforces this exploration through the identification of structural relationships for systematic data extraction [latex]\mathcal{E}1[/latex], ultimately consolidating findings into structured, searchable databases [latex]\mathcal{E}2[/latex].](https://arxiv.org/html/2603.18447v1/x1.png)
Researchers have developed a new framework to automatically transform unstructured information found across the internet into structured, queryable databases.
Tuning Diffusion Models Is Simpler Than You Think
New research demonstrates a remarkably data-efficient method for aligning powerful generative models with human preferences using a streamlined fine-tuning process.
AI Takes the Stand: Teaching Cybersecurity with an Intelligent Tutor
A new agentic AI system, the CyberJustice Tutor, is demonstrating a promising approach to cybersecurity education by adapting to individual student needs and verifying its knowledge.
Unlocking Scientific Insights with Time-Series Foundation Models
![Distillation of the STEP encoder from varying teacher models yields differing performance, as quantified by [latex]F_1[/latex] scores on downstream scientific time series tasks.](https://arxiv.org/html/2603.18688v1/x3.png)
A new framework, STEP, is enabling more effective analysis of complex scientific data by pretraining encoders across diverse domains.
Seeing Clearly: Boosting Visual SLAM in Difficult Conditions
A new review assesses how different visual feature extraction methods impact the robustness and accuracy of LiDAR-inertial-visual odometry systems when faced with challenging environments.
Closing the Reality Gap: Generative Worlds Boost Robot Learning
New research demonstrates that training robots in dynamically generated 3D environments dramatically improves their ability to perform tasks in the real world.
Decoding the Molecular World: From Mass Spectra to Structures
A new deep learning approach accurately predicts molecular structures directly from mass spectrometry data, promising faster and more efficient chemical analysis.