鹭羽 发自 凹非寺 量子位 | 公众号 QbitAI 2025都有哪些AI趋势，大神 卡帕西 的年终总结，正在火爆硅谷。 6大论断，硬核又颇有启发： 新范式、新应用、新模型……回首望去，过去一年大模型带来的变革让人兴奋。 然而卡帕西大胆预言： 大模型的潜力，才刚刚挖掘10%。 一切不过是刚刚开始…… 2025LLM年度回顾 为什么卡帕西认为大模型潜力只挖掘了10%？ 一方面展现出强大的推理能力，另一方面也暴露出潜在的理解缺陷 ，既让人兴奋又让人谨慎，具体包括： RLVR （可验证奖励强化学习） 成为训练新阶段 大模型不应被类比为动物智能 Cursor展现了大模型应用的Next Level Claude Code加速端侧智能体普及 Vibe Coding将重塑软件行业 Nano Banana重塑人机交互 RLVR成为训练新阶段 在年初之前，全世界的大模型都基本遵循以下训练范式： 而到了2025年，RLVR开始加入其中。 模型通过在可自动验证的奖励环境中进行强化学习训练，会自发地形成推理策略，比如将问题分解为中间计算、循环计算等，具体可参考 DeepSeek R1 。 而这些策略如果用旧范式其实极难实现，因为大模 ...

Core Insights - The discussion at the 2025 World Internet Conference highlighted the limitations of current AI technologies, particularly in their ability to generalize across different domains and perform simple tasks effectively [1] - The concept of "jagged intelligence" was introduced, indicating that while AI can excel in complex areas, it still struggles with tasks that humans find straightforward [1] - The need for AI to develop stable generalization learning algorithms and establish more connections with the real world was emphasized, suggesting a path towards more human-like learning capabilities [1] Group 1 - The leaders of the "Hangzhou Six Little Dragons" gathered for a dialogue at the conference, indicating a collaborative effort among key players in the tech industry [1] - Chen Deli, a senior researcher at DeepSeek, pointed out the current inadequacies of AI in self-iteration and evolution compared to human learning processes [1] - The discussion included the importance of multimodal and embodied intelligence to enhance AI's learning in real-world environments [1] Group 2 - Looking ahead 10 to 20 years, there is optimism about the potential for achieving Artificial General Intelligence (AGI), as technological advancements often accelerate over time [2] - The example of ChatGPT's rapid improvement in solving mathematical problems illustrates the potential for significant breakthroughs in AI capabilities [2] - The sentiment expressed is that the current phase of the AI revolution is still in its early stages, with expectations for future advancements [2]

Core Insights - The interview with Demis Hassabis, CEO of Google DeepMind, discusses the evolution of AI technology and its future trends, particularly focusing on the development of general artificial intelligence (AGI) and the significance of world models like Genie 3 [2][3]. Group 1: Genie 3 and World Models - Genie 3 is a product of multiple research branches at DeepMind, aimed at creating a "world model" that helps AI understand the physical world, including physical structures, material properties, fluid dynamics, and biological behaviors [3]. - The development of AI has transitioned from specialized intelligence to more comprehensive models, with a focus on understanding the physical world as a foundation for AGI [3][4]. - Genie 3 can generate consistent virtual environments, maintaining the state of the scene when users return, which demonstrates its understanding of the world's operational logic [4]. Group 2: Game Arena and AGI Evaluation - Google DeepMind has partnered with Kaggle to launch Game Arena, a new testing platform designed to evaluate the progress of AGI by allowing models to play various games and test their capabilities [6]. - Game Arena provides a pure testing environment with objective performance metrics, allowing for automatic adjustment of game difficulty as AI capabilities improve [9]. - The platform aims to create a comprehensive assessment of AI's general capabilities across multiple domains, ultimately enabling AI systems to invent and teach new games to each other [9][10]. Group 3: Challenges in AGI Development - Current AI systems exhibit inconsistent performance, being capable in some areas while failing in simpler tasks, which poses a significant barrier to AGI development [7]. - There is a need for more challenging and diverse benchmarks that encompass understanding of the physical world, intuitive physics, and safety features [8]. - Demis emphasizes the importance of understanding human goals and translating them into useful reward functions for optimization in AGI systems [10]. Group 4: Future Directions in AI - The evolution of thinking models, such as Deep Think, represents a crucial direction for AI, focusing on reasoning, planning, and optimization through iterative processes [12]. - The transition from weight models to complete systems is highlighted, where modern AI can integrate tool usage, planning, and reasoning capabilities for more complex functionalities [13].