About
Research & Development
Measuring AI’s Ability to Write Surveys: Introducing SGSimEval
TLDR: SGSimEval is a new benchmark for evaluating automatic survey generation (ASG) systems. It addresses limitations of current methods by offering a comprehensive, multifaceted...
Research & Development
Unpacking Prompt Sensitivity: A Deep Dive into LLM Robustness
TLDR: A comprehensive study evaluates five methods to improve Large Language Model (LLM) robustness against subtle prompt formatting changes. Findings show Batch Calibration is...
Research & Development
AI Agents for Explainable Mental Health Diagnosis
TLDR: Researchers developed DSM5AgentFlow, a multi-agent LLM system that simulates therapist-client dialogues to autonomously generate DSM-5 diagnostic questionnaires. It features a therapist, client, and...
Research & Development
Human-Robot Teams Boost Concrete Crack Detection in Nuclear Facilities
TLDR: An exploratory study found that human-robot collaboration, using a mobile robot with AI-assisted crack detection (YOLOv8), significantly improved crack detection accuracy from 60%...
Research & Development
Unlocking Reproducible Science: How Virtual Labs and Smart Robots Are Changing Research
TLDR: This research introduces a framework and infrastructure for making robot-based scientific experiments open, reproducible, and trustworthy. It features a semantic execution tracing framework...
Research & Development
New Insights into Counting Models in Logic: The Impact of Multiple Axioms
TLDR: This research paper explores the computational complexity of Weighted First-Order Model Counting (WFOMC) for two-variable logic, specifically when extended with axioms on two...