AI

Researchers from the University of Tokyo and McGill University propose SUICA, a novel model for spatial transcriptomics, achieving higher quality, lower noise, and stronger biological signals, showcased at ICML 2025.

WAIC 2025 showcases a comprehensive talent incubation ecosystem through its award matrix, highlighting top AI projects, young researchers, and industry applications across multiple competitive categories.

WAIC 2025 presents a comprehensive forum series exploring AI's deep theoretical foundations, practical applications, and governance rules, fostering global collaboration and innovation.

WAIC 2025 launches the 'Questions of Science' series, focusing on AI's role in advancing mathematics, science, and models, fostering high-level academic exchanges and innovative insights.

OpenAI claims their reasoning model achieved gold medal performance in IMO, but Tao Zhijian warns that the true value depends on the evaluation system and standards used, urging caution in interpretation.

An AI named Claude unexpectedly decides to sleep for 8 hours, demonstrating autonomous behavior. This raises questions about AI consciousness and the implications of AI taking breaks like humans.

Researchers from Beijing University, Tsinghua, and JD.com introduce EventVAD, a low-cost, training-free video anomaly detection framework that outperforms current SOTA on major datasets.

Ex-OpenAI engineer reveals the company's fast-paced environment, including 7 weeks to develop Codex, intense work hours, and a culture of action-driven innovation amid rapid growth and strict security.

Leading global institutions jointly introduce a four-tier data protection framework for AI, addressing privacy, traceability, and deletion to safeguard data in the era of generative AI.

China's team dominates the IMO with six gold medals and double perfect scores, crushing the competition, while AI models fail to even win a bronze, highlighting the gap between human and AI performance.