Meta的首席人工智能科学家Yann LeCun近日在社交平台X（原Twitter）上对OpenAI新发布的视频生成模型Sora提出了质疑，直言其“注定要失败”。Sora自发布以来，因其在人工智能领域的创新性而备受业界关注。然而，LeCun对此持有不同观点，他认为通过生成像素来模拟世界的方法类似于过去的“通过合成进行分析”，这一做法既不经济又可能走向失败。
LeCun指出，这种技术路线可能会消耗大量资源，而且回避了更有效的AI学习策略。他强调，当前AI研究应更注重理解和模拟现实世界的复杂性，而非仅仅在像素层面进行模拟。这一评论引发了业界对于AI发展方向的深入讨论，人们开始反思是否应该重新评估基于像素生成的AI模型的价值和效率。
尽管LeCun的评论可能对Sora的未来前景蒙上了一层阴影，但OpenAI尚未对这一批评作出官方回应。作为AI领域的重量级人物，LeCun的观点往往能影响业界的风向，因此Sora的技术路径是否真的如他所言“注定失败”，还需时间来验证。同时，这也为AI研究领域提供了一次反思和探讨新方法的机会。
英语如下：
News Title: “Meta’s Chief AI Scientist LeCun Critiques OpenAI’s New Model Sora as a ‘Doomed Resource Waste’
Keywords: Meta, LeCun, Sora
News Content: Yann LeCun, Meta’s chief artificial intelligence scientist, recently voiced skepticism about OpenAI’s newly released video generation model, Sora, on social platform X (formerly Twitter), labeling it “doomed to fail.” Since its launch, Sora has garnered significant industry attention for its innovations in AI. However, LeCun holds a different perspective, arguing that the approach of generating pixels to simulate the world is akin to past “synthetic analysis,” which he deems both inefficient and prone to failure.
LeCun pointed out that this technological path might consume substantial resources and sidesteps more efficient AI learning strategies. He emphasized that current AI research should focus more on understanding and simulating the complexity of the real world, rather than mere pixel-level simulation. This commentary has sparked in-depth discussions within the industry about the direction of AI development, prompting reflections on whether the value and efficiency of AI models based on pixel generation should be reassessed.
While LeCun’s comments might cast a shadow over Sora’s future prospects, OpenAI has yet to respond officially to the criticism. As a prominent figure in the AI field, LeCun’s opinions often sway industry trends, suggesting that Sora’s technological trajectory might indeed be “doomed” as he claims. Time will be the deciding factor. In the meantime, this episode offers an opportunity for the AI research community to reflect and explore new methodologies.
【来源】https://www.ithome.com/0/751/676.htm
Views: 1