Search papers, labs, and topics across Lattice.
School of Computer Science, Beijing University of Posts and Telecommunications
3
0
6
4
Today's best vision-language models are surprisingly bad at reading scientific figures, failing to match expert-level reasoning on a new benchmark of experimental images.
LLMs can now generate research roadmaps that are 8% better and 84% faster than human experts, thanks to a novel multi-agent system.
Even GPT-5.1 struggles to distinguish AI-generated academic images from real ones, achieving only 48.8% accuracy, revealing a significant gap between generative and forensic AI capabilities.
