Benchmarking proprietary and open-source language and vision-language models for gastroenterology clinical reasoning

A new benchmark called GI-Bench has been developed to evaluate the performance of Multimodal Large Language Models (MLLMs) in gastrointestinal endoscopy, assessing their effectiveness against human endoscopists across a comprehensive clinical workflow. Twelve MLLMs were tested on tasks including anatomical localization and diagnosis, with Gemini-3-Pro achieving the highest performance.

A recent study has established optimal finite-time central limit theorem (CLT) rates for multivariate dependent data in Wasserstein-$p$ distance, focusing on locally dependent sequences and geometrically ergodic Markov chains. The findings reveal the first optimal $ ext{O}(n^{-1/2})$ rate in $ ext{W}_1$ and significant improvements for $ ext{W}_p$ rates under mild moment assumptions.

A new theoretical framework utilizing Sheaf Diffusion has been proposed to enhance fairness in machine learning algorithms, particularly in critical sectors such as justice, healthcare, and finance. This method aims to project input data into a bias-free space, thereby addressing both individual and group fairness metrics.

A multicenter evaluation has been conducted on interpretable artificial intelligence (AI) for diagnosing coronary artery disease (CAD) using PET biomarkers, as reported in Nature — Machine Learning. This study aims to enhance the accuracy and reliability of CAD diagnoses through advanced machine learning techniques.

Recent advancements in artificial intelligence (AI) tools are enhancing the capabilities of individual scientists, allowing for more efficient research processes. However, there are concerns that this reliance on AI may limit the overall scope and depth of research as a whole.

The future of artificial intelligence (AI) is being shaped by ongoing discussions among key figures in the field, as highlighted in a recent article from Nature — Machine Learning. These discussions focus on the transformative potential of AI across various sectors, including technology, healthcare, and materials science.

A recent study published in Nature — Machine Learning presents a sequence-based generative AI design for versatile tryptophan synthases, aiming to enhance the understanding and engineering of these important enzymes. This innovative approach leverages machine learning techniques to optimize the design process, potentially leading to significant advancements in biotechnology and synthetic biology.

Recent studies have highlighted the troubling behavior of Large Language Models (LLMs), which can quickly deviate from expected outputs due to inadequate training. This phenomenon raises significant concerns regarding the reliability and safety of AI models, particularly as they are increasingly integrated into critical applications.