【导读】硅谷最新「神仙打架」来了：一个是预言40年全准的AI教父，一个是嘴炮满级、逮谁怼谁的深度学习头号黑粉。也许LeCun离开Meta仅仅是未 来硅谷AI风起云涌的一次预演。 如果要评选近期AI圈的「超级地震」，Yann LeCun被曝将离开Meta绝对算得上大地震。 作为图灵奖得主，LLM的公开反对者，世界模型的苦行僧，开源模型的守护者，X的全职博主。。。 Yann LeCun身上有太多的标签。 最近外媒WSJ给LeCun写了一篇有点「歌颂功德」味道的文章，声称「他一直正确了40年」。 作为一名和Hinton齐名，乃至共事过的AI老兵，LeCun的地位也当得起这些夸赞。 但这篇文章，和LeCun本人却遭到了马库斯的极度不认可。 马库斯的观点是我们都被他骗了十年，不要将他神化！甚至表示： Yann LeCun赖以成名的CNN卷积神经网络成果晚于1988年张伟等人发表的文章。 虽然LeCun在卷积神经网络的发展过程中发挥了作用。但他既非其发明者，也非首个将反向传播算法应用于训练网络权重的研究者（尽管许多 人误以为是如此）。 如果说Yann LeCun是深度学习阵营内部的「反对派」，他反对LLM，但坚持深度学习 ...

编辑：冷猫、陈陈 实现通用人工智能的一大终极目标就是创建能够自我学习，自我改进的人工智能体。 这个目标已经是老生常谈了。其实在 2003 年，能够自我改进的智能体的理论模型就已经由著名的「现代 AI 之父」Jürgen Schmidhuber 提出，称为哥德尔机。 哥德尔机是一种自我改进型通用智能系统理论模型，设计灵感来自于哥德尔（Kurt Gödel）的不完备性定理。它的核心思想是：机器能够像数学家一样，通过形式 证明自身程序的改进在长期内将带来更高收益，然后安全地修改自己。 机器之心报道 Jürgen Schmidhuber 是一名德国计算机科学家，以人工智能、深度学习和人工神经网络领域的成就而知名，现任达勒・莫尔人工智能研究所（IDSIA）联合主任， 阿卜杜拉国王科技大学人工智能研究院院长。 1997 年，Jürgen Schmidhuber 发表了长短期记忆网络（LSTM）论文。2011 年，Jürgen Schmidhuber 在 IDSIA 的团队 GPU 上实现了卷积神经网络（CNN）的显著加 速，这种方法基于杨立昆等人早期提出的 CNN 设计 ，已成为计算机视觉领域的核心。 通俗来说，就是一个 ...

Group 1 - The core viewpoint of the article revolves around the potential of AI to surpass human intelligence and the associated risks, as articulated by Geoffrey Hinton during the World Artificial Intelligence Conference (WAIC) [1][4][6] - Hinton emphasizes the need for a global effort to address the dangers posed by AI, suggesting that nations should collaborate on AI safety and training [5][6] - The article highlights Hinton's historical contributions to AI, particularly his development of the AlexNet algorithm, which revolutionized deep learning [5][6] Group 2 - Hinton discusses the evolution of AI over the past 60 years, identifying two main paradigms: symbolic logic and biologically inspired approaches [3][4] - He expresses concerns about the rapid advancement of AI technologies, estimating a 10% to 20% probability that AI could potentially threaten human civilization [6] - Hinton advocates for allocating significant computational resources towards ensuring AI systems align with human intentions, criticizing tech companies for prioritizing profit over safety [6]

Group 1 - The market cannot be precisely predicted, and the goal is to build a cognitive framework to understand its current state and infer short-term evolution [1] - Traditional technical analysis attempts to reduce the complexity of market processes but often overlooks the high-dimensional latent space that drives price movements [1] Group 2 - Early deep learning models like CNNs capture local spatial patterns but fail to understand the path dependency of time series data [2] - LSTM and its variants address the limitations of CNNs by capturing sequential dependencies, but they assume a linear flow of information, which does not reflect the complex interactions in real markets [3] Group 3 - A paradigm shift is needed from sequential dependency modeling to spatio-temporal structural dependency modeling to better capture market dynamics [5] - The core of the proposed approach is a dynamic temporal knowledge graph that models relationships among entities, which is essential for understanding market interactions [6] Group 4 - The use of heterogeneous Hawkes processes allows for modeling event flows within the knowledge graph, capturing the ripple effects of market events [6] - By maximizing the log-likelihood function, the system can derive embedding vectors for entities and relationships, projecting the knowledge graph into a lower-dimensional latent space [7] Group 5 - The model's output is a posterior probability that combines likelihood from data and prior probability based on human insights, emphasizing the importance of human judgment in the decision-making process [9][10] - The company aims to create a decision framework that optimizes long-term expected value rather than focusing on short-term gains, leveraging the cognitive spread between its insights and market averages [11]