Core Insights - Shanghai Jiao Tong University (SJTU) and Xuhui District Government hosted the "AI Future Development Forum: Super Intelligence, Boundless Co-creation" as part of the 2025 World Artificial Intelligence Conference, showcasing significant AI research achievements and the potential for large-scale applications in various industries [1][3] Group 1: AI Research Achievements - SJTU introduced the Innovator scientific base model and the SciMaster research intelligence system, which demonstrate global leadership in scientific multimodal understanding and reasoning capabilities, serving millions of users through an open-source research platform [2] - The first-ever end-side native sparse large model was presented, designed to match heterogeneous computing power and memory constraints of end devices, enabling efficient deployment of a 10 billion parameter model without the need for internet connectivity [2] Group 2: AI Future Development Initiatives - The forum launched the AI Future Fund, initiated by SJTU's AI Institute and supported by outstanding alumni, focusing on connecting global young AI talents through seed research and innovation programs [3] - A collaborative research initiative with Nature was announced to address ten future AI questions, aiming to identify forward-looking and interdisciplinary topics in the AI field, with results to be published during SJTU's 130th anniversary in 2026 [5]

Core Viewpoint - Shanghai Jiao Tong University, Shanghai Algorithm Innovation Institute, and DeepMind Technology jointly launched the world's first general-purpose scientific research AI, SciMaster, based on the Innovator model [1] Group 1 - SciMaster integrates numerous specialized scientific tools and can generate "in-depth research reports" [1] - The AI supports a thinking chain editing function, allowing researchers to actively intervene in SciMaster's execution logic [1] - Researchers can modify task logic and content to achieve more accurate and reasonable research needs [1] Group 2 - SciMaster is now connected to the DeepModeling open-source community [1]