最新资讯
dev.to 为 CI/CD 构建 AI 质量门禁 dev.to Kakao 发布面向 AI Agent 的决策模型 Kanana-2-v-decide computerworld.com Microsoft 对 Copilot Chat 企业用户扩大未公开的使用限制 pub.towardsai.net Google 发布 Gemini 4 Carbon:主打网络安全能力的前沿模型 gradientflow.com METR 研究:AI 正在加速构建 AI 的团队本身 www.infoq.com GitLab 19.5 将 Agentic 开发纳入 DevOps 平台的治理体系 aiacceleratorinstitute.com 研究:AI Agent 的复杂流程编排可能反而是问题所在 pub.towardsai.net MongoDB 推出 Vercel AI SDK 集成:为 AI 应用提供持久记忆 thenewstack.io Google 捐出被 Anthropic 和 OpenAI 使用的代码沙箱 gVisor,押注 AI Agent 基础设施开源 dev.to 在信任 AI 生成的测试用例前,如何逐条复核 unite.ai Franklin Templeton 构建 AI 研究工作流以揭示风险 techxplore.com AI 助力寻找更优的钛合金 3D 打印工艺参数
多元动态
登录 注册
大模型 www.aibase.com ·

DeepMind 成立新研究院,四篇文章聚焦 AGI 关键议题

原文 我的收藏

Google 与 Google DeepMind 研究人员于 9 月 16 日成立 DeepMind Institute,推动围绕通用人工智能(AGI)的跨学科讨论,董事会包括 DeepMind 联合创始人 Shane Legg、Google 高管 James Manyika 和 Google DeepMind 主席 Demis Hassabis,Legg 任执行编辑。首批四篇文章分别讨论 AGI 经济政策、模型推理透明度、人类繁荣原则和前沿 AI 评估框架。Rohin Shah 与 Anca Dragan 提出应关注模型“透明窗口”缩小带来的安全风险,可通过限制“不可见串行深度”或要求低透明度系统证明其仍保持等效可控性来应对。Hassabis 建议建立由美国主导的前沿 AI 标准组织,开发者可在模型发布前最多 30 天自愿提交评估;系统成熟后,通过测试可能成为在美国部署前沿模型的必要条件。评估机构最终将采用独立、非公开的“保密”测试,以降低模型针对已知指标优化的可能,并可按风险等级进一步加强措施。

AI推荐理由 内容为 DeepMind 成立新研究院并发布四篇 AGI 相关文章的转述,包含具体人物(Shane Legg、James Manyika、Demis Hassabis)、具体议题(AGI 经济政策、推理透明度、人类繁荣原则、前沿 AI 评估框架)以及 Rohin Shah、Anca Dragan 关于模型“透明窗口”缩小带来安全风险、限制“不可见串行深度”或要求低透明度系统证明可控性的主张,还提到 Hassabis 建议建立由美国主导的前沿 AI 标准组织、开发者可在发布前最多 30 天自愿提交评估、成熟后通过测试或成为部署必要条件、最终采用独立非公开“保密”测试。这些是治理与评估机制层面的具体新信息,对 AI 安全与政策方向有参考价值。但来源为 AIbase 的二次转述,非官方原文,正文未提供四篇文章链接或原文细节,部分表述(如“可能成为必要条件”)本身带有不确定性,需保留。

72

正文

DeepMind Establishes New Institute to Address Key Issues in AGI AIbase基地 Published inAI News · 3 min read · Sep 18, 2026 6 On September 16, Google and Google DeepMind researchers established the DeepMind Institute to promote interdisciplinary discussions around general artificial intelligence (AGI). The institute's board includes Shane Legg, co-founder of DeepMind, Google executive James Manyika, and chairman of Google DeepMind, Demis Hassabis, with Legg serving as executive editor. The institute emphasizes presenting different perspectives from Google, DeepMind, and the global research community on AGI issues. The institute's first four articles discuss AGI economic policy, model reasoning transparency, principles of human flourishing, and a frontier AI evaluation framework. Rohin Shah and Anca Dragan proposed that attention should be paid to the security risks caused by the shrinking "transparent window" of models, which can be addressed by limiting "invisible serial depth" or requiring low transparency systems to prove they still maintain equivalent controllability. Hassabis also suggested establishing a frontier AI standards organization led by the United States, where developers could voluntarily submit assessments up to 30 days before model release; once the system matures, passing the test may become a necessary condition for deploying frontier models in the US. The evaluation organization will eventually adopt independent, non-public "confidential" tests to reduce the possibility of models optimizing for known metrics, and further measures can be strengthened based on risk levels. This move shows that discussions on AGI are extending further from capability development to specific topics such as reasoning transparency, external evaluation, and governance mechanisms. GoogleDeepMindAGIAI Terms This article is from AIbase Daily Scan to view Welcome to the [AI Daily] column! This is your daily guide to exploring the world of artificial intelligence. Every day, we present you with hot topics in the AI field, focusing on developers, helping you understand technical trends, and learning about innovative AI product applications. —— Created by the AIbase Daily Team © Copyright AIbase Base 2024, Click to View Source -

来源报道 · 1 篇 · 以下链接均前往原始媒体

← 返回精选动态 原文

更多内容