人工智能开源

Search documents
信通院王爱华:开闭源各有特点,开源对AI普惠起到更大作用
Bei Ke Cai Jing· 2025-07-10 07:55
Core Viewpoint - The development of an open-source ecosystem in artificial intelligence (AI) is crucial for enhancing China's AI industry, with open-source large models playing a significant role as AI becomes a global public good and enters a stage of inclusivity [1][3]. Group 1: Open Source Ecosystem Development - Beijing has the foundational capabilities to build an "open-source capital," hosting over half of the country's open-source startups across key areas such as foundational platforms, large models, and robotics [3]. - The city possesses the largest information software industry in the country, with a robust industrial chain that supports open-source development [3]. - Continuous improvement of top-level design encourages enterprises to actively participate in international open-source projects and supports the establishment of open-source organizations in Beijing [3]. Group 2: Recommendations for AI Open Source Ecosystem - It is recommended to strategically position in frontier areas such as intelligent agents and embodied intelligence to accelerate technological breakthroughs and ecosystem development [3]. - Creating a favorable open-source culture is essential, encouraging institutions to practice open-source principles in the AI field [3]. - Integration of inclusive computing resources and the establishment of a joint operation mechanism are necessary for the ecosystem [3]. - Strengthening communication and collaboration within existing AI open-source communities is vital to create a neutral and authoritative platform [3]. Group 3: AI Open Source Community - The AI open-source community has become a key component for the practical application of large models, exemplified by the Jingzhi community initiated by the China Academy of Information and Communications Technology [5]. - As of now, the Jingzhi community has over 2,800 open-source models covering areas such as computer vision, natural language processing, speech processing, and multimodal applications [5]. - The community has aggregated more than 380 datasets, including text, images, and audio-visual data, and provides over 500 petaflops of computing power from resources like NVIDIA, Huawei Ascend, and Baidu Kunlun [5].
DeepSeek引发全球AI价格战,OpenAI开源能否重塑行业格局?
Sou Hu Cai Jing· 2025-05-26 12:08
作者|杨睿琪 编辑|刘毓坤 AI 行业正迎来前所未有的剧变。DeepSeek-R1 凭借大幅降低的推理成本,点燃全球 AI 价格战导火索, 而 OpenAI 近期宣布开源新模型并推出 "云端接力" 功能,将 "闭源 VS 开源" 之争推向高潮。 "普及度可能不是最准确的词。"OpenAI前商业化负责人Zack Kass强调,"但行业已达成共识——开放和 封闭两种策略在当下缺一不可。" 这场技术与商业的博弈背后,究竟藏着怎样的行业趋势?OpenAI 前商业化负责人 Zack Kass 在接受凤 凰网科技采访时,给出了深度解读。 "前沿模型的价值是有限的",Zack Kass一针见血的指出。之所以有此判断,是源于今天开源社区展现 出惊人的技术复现能力,过去需要多年完成的技术突破,现在仅需 6 个月就能实现。这似乎意味着,闭 源商业模式曾引以为傲的技术护城河,正在被开源浪潮不断冲刷。 OpenAI的开源动作被解读为应对市场竞争的防御性举措。此前DeepSeek-R1大幅降低推理成本引发全球 价格战,"AI将和互联网一样廉价"的预言正在应验。"基础研究成本骤降是科技领域最美好的事。"Zack Kass指出,"虽然短期 ...
超越DeepSeek,中国开源“集团军”重塑全球AI生态
Guan Cha Zhe Wang· 2025-04-27 12:57
中国开源,正在形成集团化作战的优势。DeepSeek和阿里Qwen等基础模型撑起了中国开源 的上限,而更多的中小企业正在它们的基础上,不断推出体量更小却能力更强的垂直模型, 让今年以来中国大模型更新的速度不断加快,不断带来新的惊喜。而在美国仍以闭源为主的 背景下,中国企业拥抱开源的做法充分展现了中国的技术自信,也开辟出一条技术普惠与全 球共生的新路径,持续开拓海外市场,代表着全球AI技术从"单极霸权"转向"多极共生"。 中国开源,正形成合力 2月初,当中国开源大模型DeepSeek在全球140个国家和地区的应用市场下载排行榜同时登顶,OpenAI 却公然在媒体上指责,DeepSeek未经许可使用了ChatGPT的蒸馏数据。 这样的指责非但没能为OpenAI"挽尊",却招来全球科研从业者的"群嘲"。 如今,又一个把蒸馏buff叠满的选手出现了。 昆仑万维这份成果,充分展示了开源最重要的意义——不仅仅是让给用户提供一个免费可用的产品,更 能够让更多开发者站在前人的肩膀上,快速、低成本地推动技术继续前进。 事实上,就在去年业界还在讨论大模型预训练遭遇瓶颈的背景下,今年以来,中国大模型的迭代速度却 仍在加快,并且越来 ...
OpenAI 的 75 封内部邮件,一堂硅谷创业课
晚点LatePost· 2024-12-24 12:53
3 万字全文翻译,被披露的不只是内斗。 文丨贺乾明 翻译、校对丨GPT-4o、Claude-3.5-Sonnet、贺乾明 编辑丨 黄俊杰 "创业公司的内部全都是车祸现场。只是有的你能在媒体上看到,有的看不到。" 山姆·阿尔特曼(Sam Altman)的导师,硅谷孵化器 YC 创始人保罗·格雷厄姆 曾这么总结他看过的千百家创业公司。 得益于阿尔特曼和特斯拉 CEO 埃隆·马斯克的争斗和诉讼,现在我们也能看到 OpenAI 早年的真实样子。双方各自公布了数批共 75 封内部邮件以及短信记 录,时间跨度从 2015 年 OpenAI 筹备到 2019 年组建营利实体,展示了一群硅谷名流和天才 AI 研究者如何因为理想聚在一起,又如何随着 OpenAI 的发展争 权夺利。 这些长达 3 万多字的内部记录也像是一堂 OpenAI 出品的创业课,涉及这家全球最大的 AI 创业公司早期如何讲故事、组建精英阵容,到薪酬设计、股权分配 的方方面面;从低价与 Google 争夺人才到与微软谈判合作,甚至还包括曾考虑过的加密货币融资方案。而且,我们还能看到首席科学家伊利亚·苏茨克维 (Ilya Sutskever) 如何写双周报, ...