Researchers Tried Letting AI Do Science. It Failed⭐9

A new study found AI failed at autonomous scientific discovery, generating nonsensical results and misinterpreting data, underscoring human oversight's necessit...

Computer Science > Computation and Language⭐9

Authors validate chatbots at scale using customer digital twin simulations, enabling robust testing with synthetic conversations replicating real user behaviors...

Computer Science > Computer Vision and Pattern Recognition⭐9

Track-leakage-free hold-out self-validation protocol for photogrammetric reconstruction reliability without ground truth, analyzing sensitivity and limits.

Computer Science > Artificial Intelligence⭐9

Explore a novel approach to visual counterfactual explanations using diffusion models, offering transparent AI insights by identifying minimal concept-based ima...

Computer Science > Computer Vision and Pattern Recognition⭐9

FogDrive: a multi-modal synthetic driving dataset for perception under graded fog conditions, enabling robust autonomous driving sensor fusion evaluation.

Computer Science > Computer Vision and Pattern Recognition⭐9

DINOv3-MIL uses self-supervised foundation model patch tokens to enable per-kidney multi-label tumour and cyst detection from the KiTS23 dataset.

Computer Science > Computation and Language⭐9

A systematic review compares bag-of-words and BiomedBERT for automated MeSH classification, finding evaluation design influences the expert-vs-auto performance ...

Computer Science > Artificial Intelligence⭐9

UP-NRPA paper introduces user portrait-based nested rollout policy adaptation for improving LLM planning in goal-oriented dialogue systems.