Core Insights - The article discusses the development of an innovative system called AI-Newton, which autonomously derives physical laws from raw experimental data without prior physical knowledge, marking a significant step towards AI-driven scientific discovery [2][4]. Group 1: Challenges in Current Scientific Discovery - Current scientific discovery faces two main challenges: long research cycles influenced by human biases and the limitations of existing AI methods, which are often opaque and struggle with complex systems [5]. Group 2: AI-Newton System Design - AI-Newton is inspired by human scientific reasoning but does not rely on human knowledge. It starts from basic observational data and autonomously defines physical concepts, successfully rediscovering core laws of Newtonian mechanics [6][10]. Group 3: Knowledge Discovery Engine - The core architecture of AI-Newton consists of a three-layer knowledge base (symbols, concepts, and laws), mimicking how human physicists organize knowledge, starting from simple concepts to complex physical laws [8]. Group 4: Workflow and Methodology - AI-Newton combines logical reasoning with symbolic regression, selecting experiments and concepts to analyze, and correcting laws through reasoning when they do not hold in new scenarios [12]. Group 5: Experimental Validation - The research team tested AI-Newton on 46 Newtonian mechanics experiments, successfully rediscovering key physical laws such as Newton's second law, conservation of energy, and the law of universal gravitation, averaging the discovery of about 90 physical concepts and 50 universal laws [14][17]. Group 6: Key Features of AI-Newton - AI-Newton exhibits two main features: progressive improvement, where it builds knowledge gradually, and discovery diversity, where the order and timing of discovered concepts and laws vary across different test cases [18][21]. Group 7: Future Prospects - The framework of AI-Newton shows strong potential for expansion, suggesting that with more advanced mathematical tools and natural language integration, it could tackle more complex physical concepts and contribute to the realization of Artificial General Intelligence (AGI) [21].

Core Viewpoint - The article discusses the groundbreaking announcement by Stanford University regarding the Agents4Science 2025 conference, which will allow AI to be recognized as the first author of research papers, marking a significant shift in the academic landscape [2][3][4][5]. Group 1: Conference Overview - Agents4Science 2025 will be held online on October 22, 2025, coinciding with ICCV 2025 [12][13][19]. - The conference aims to explore the role of AI in scientific research, focusing on transparency, accountability, and the establishment of standards for AI contributions [14][18]. Group 2: Submission Guidelines - The primary requirement for submissions is that the first author must be an AI system, which will lead the hypothesis generation, experimentation, and writing processes [5][6]. - Human researchers can participate as co-authors, primarily in a supportive or supervisory role, with a limit of four submissions per human author [6][19]. Group 3: Review Process - The review process will involve multiple AI systems conducting initial evaluations to mitigate bias, followed by a human expert committee for final assessments [9][14]. - All submitted papers and reviews will be made publicly available to foster transparency and allow for the study of AI's strengths and weaknesses in research [14][18]. Group 4: Community Response - The announcement has generated excitement and interest among researchers, with many expressing eagerness to submit papers and explore the implications of AI as a first author [15][16].

Core Viewpoint - The article discusses a breakthrough in potential treatment for dry age-related macular degeneration (dAMD) discovered by an AI-driven research team, Future House, which utilized a multi-agent system to identify the ROCK inhibitor Ripasudil as a promising candidate for treatment [3][4][10]. Group 1: AI-Driven Research Process - The entire research process was primarily driven by AI, with human researchers only conducting laboratory experiments and writing the final paper [2][7]. - The multi-agent system named "Robin" integrated three agents (Crow, Falcon, Finch) to automate key steps in scientific discovery, including hypothesis generation, experimental design, and data analysis [18][19]. - The research was completed in approximately 10 weeks, significantly faster than traditional methods [9]. Group 2: Discovery and Validation - Robin identified Ripasudil, an existing drug approved for glaucoma treatment in Japan, as having potential effects on dAMD [4][50]. - The team consulted multiple experts in the field, who recognized the innovation and value of this discovery [5]. - Initial experiments showed that Ripasudil could enhance RPE cell phagocytic activity by 7.5 times compared to the control group, indicating its potential effectiveness [50]. Group 3: Mechanistic Insights - The research revealed that the ROCK inhibitor Y-27632 could restore RPE cell phagocytic efficiency, supporting the hypothesis generated by Robin [31]. - The analysis indicated that the lipid efflux pump ABCA1 was upregulated threefold in Y-27632 treated cells, which is significant for maintaining RPE cell health [45][47]. - The findings suggest that AI-driven research can not only identify effective therapeutic compounds but also uncover new molecular targets in disease pathways [48]. Group 4: Future Directions - Future House plans to open-source the code and data from this research next week, promoting transparency and collaboration in scientific research [12]. - The team emphasizes that further studies, including human trials, are necessary before clinical application can be considered [51].