An analysis of how AI-generated research papers are measured on arXiv, highlighting current methods and their limitations.
How We Measured AI Writing Across arXiv, And Where The Measurement Breaks
Browsing Category
AI & Tooling
62 posts
AI Advice Made People 3X Less Accurate But 2X Confident, Researchers Found
Research shows people become less accurate but more confident when following AI advice, raising questions about reliance on AI guidance.
GPT-5.6 Used A Prompt To Close A 30-Year Gap In Convex Optimization
GPT-5.6 achieved a breakthrough in convex optimization by closing a 30-year research gap through a novel prompt-based approach, confirmed by researchers.
Static Search Trees: 40X Faster Than Binary Search (2024)
New research reveals static search trees are up to 40 times faster than traditional binary search, promising significant performance gains for data retrieval.
Real AI Tests Show Business Skills Trump Chatbots — Only Two Models Conquer Crisis Week
Real AI tests reveal that closing deals and making decisive actions matter most — not just chat quality. Only two models excelled in a live company crisis experiment.
Corvus ISR Publishes Synthetic Benchmark Showing Tracker Improvements
The published matrix — every row reproducible. Source: corvusisr.com/benchmark Corvus ISR’s latest…
Detecting LLM-Generated Texts With “Classical” Machine Learning
Researchers develop a method to identify texts produced by large language models using classical machine learning techniques, enhancing detection capabilities.
Signature design move: A Look Inside “TERRANE — Ridgeline National Reserve Field Guide” (FABLE/175)
“TERRANE — Ridgeline National Reserve Field Guide” — see it live. Discover…
Can LLMs Perform Deep Technical Comprehension Of Computer Architecture Papers
This report evaluates whether large language models can perform deep technical comprehension of computer architecture research papers, highlighting confirmed capabilities and ongoing uncertainties.
LeMario: Training a JEPA World Model on Super Mario Bros
LeMario has successfully trained a JEPA World Model on Super Mario Bros, marking a significant step in AI game modeling and reinforcement learning.
