Tag with "Scientific Research"

A groundbreaking study on DeepSeek-R1, published in Nature, demonstrates enhanced reasoning in AI through reinforcement learning. Led by Professor Liang Wenfeng's team, the research shows how large language models can solve complex problems with minimal human intervention, marking a significant milestone in AI development.

InternLM has launched Intern-S1-mini, an open-source multimodal reasoning model with 8B parameters. Combining Qwen3-8B language capabilities with InternViT's visual processing, it excels in scientific data interpretation and synthesis planning. Benchmarks show superior performance in chemistry, biology, and materials science tasks.

OpenAI CEO Sam Altman forecasts AI will autonomously discover new science and humanoid robots will become commonplace within 5-10 years. He highlights AI's doctoral-level reasoning capabilities but warns of a 'superintelligence paradox' where societal impacts may lag behind technological advances. Altman also discusses OpenAI's vision for ubiquitous AI companions and Meta's aggressive talent recruitment strategies.