Cerebras CEO:人人拥有家用个人 AI 助手的未来已不远
Cerebras CEO Andrew Feldman 在 TechCrunch Disrupt 2026 上表示,人人拥有家用 AI 设备的未来已不遥远,个人 AI 助手将本地运行、部分任务回到云端,并拥有安全边界、可访问信用卡替你购物。他以 Meta 的 Muse 和 Instinct 等消费级 AI 智能体为例,并称自己已用一组智能体管理日历、查询通话对象信息。
Cerebras CEO Andrew Feldman 在 TechCrunch Disrupt 2026 上表示,人人拥有家用 AI 设备的未来已不遥远,个人 AI 助手将本地运行、部分任务回到云端,并拥有安全边界、可访问信用卡替你购物。他以 Meta 的 Muse 和 Instinct 等消费级 AI 智能体为例,并称自己已用一组智能体管理日历、查询通话对象信息。
My AI agent doesn't live in a chat window. She lives on my studio wall 🌅 Pi 5 + laser projector + No screen, no app — just her, floating on the wall, changing state as she works. We talk out loud — she listens, and answers back in her own voice. @alexandr_wang @natfriedman — your @muse SDK escaped the gadget and moved in! Still a work in progress — new senses coming..
Dyna Robotics 联合创始人兼首席科学家 Jason Ma 在 Latent Space 播客中介绍了 Dyna-2 世界动作模型,该模型基于超过一百万小时人类视频预训练,可在约 45 分钟演示数据后学会新任务。
Microsoft 合伙人研究经理 Jennifer Neville 在 Microsoft Research Podcast 中讨论评测如何推动 AI 系统突破传统 benchmark 的性能边界,以及模型在超出常规评测时出现的"意外失败"。她结合自身从数学、物理、认知科学转向计算机科学的经历,分享了使用当前 AI 系统的实用建议,并强调当结果偏离预期时应仔细审视数据。
企业 AI 的竞争前沿已从预测转向自主决策,核心难题是让预测系统在不偏离业务意图的前提下依据自身结论行动。深度学习与生成式 AI 驱动的智能分析支持实时训练,使 AI 持续演进而无需等待季度刷新,其数据来源也从规整的数值记录扩展到非结构化交互数据。Everest Group 合伙人 Vishal Gupta 称,企业已不再满足于回望式视角,"analytics"一词正让位于 AI。
Import AI 475 讨论了 AI 群体(swarm)扩展:4 个智能体的群体完成任务所需 token 总量约为单智能体的两倍,但每个智能体只需一半 token,并行运行理论上可将耗时减半。
Michael Levin 提出"Platonic mindspace"假说,认为心智是非物理模式,身体与机器只是其进入物理世界的接口,xenobots、anthrobots 等实验被视为佐证。本期 Import AI 还讨论了太空 TPU 与机器人领域等待 LLM 时刻但缺少通用算法。智谱(Zhipu)则启动了外层 RSI 循环。
One step closer to 4-8x faster Ethereum finality! It took some time and lots of tokens, but we now have a formally verified proposal for a decoupled consensus protocol in I* (a future Ethereum upgrade)! Not yet a full spec (up next), but it includes all the key consensus-relevant details to become one. Since Ethereum aspires to be live without most of the stake online, the protocol involves many more components than a normal BFT protocol, and its correctness involves much more than standard safety and liveness. Those nuanced properties are now verified! What's more, I came away convinced that all protocol design will involve AI-assisted Formal Verification in the future, both for correctness and iteration speed. The work wasn't limited to just: Design the protocol -> Formally verify it Instead, the loop became more like: Design -> Formal Model -> Find exactly what breaks and why -> Redesign it. For a fairly complicated protocol like this one, I think having the Lean model be part of the design loop played a big role in accelerating the process. A future with agents paired with formal models is a superpower for Ethereum development, because they can then use those models to find exactly where an argument breaks down, formalize counterexamples, test proposed fixes, iterate on the protocol. Many details that would slip under the radar when asking agents (and indeed, humans) can now be specified exactly and checked by the Lean kernel. This then forces agents to be more precise and lets them make verifiable progress on their own. It's been incredible to see this play out, seeing agents find gaps and propose protocol changes to fix them. In other words, autoresearch can speed up protocol design, formal verification is here to stay, and Ethereum Finality will get faster.
GitHub Copilot 提出 canvas 概念,即运行在 GitHub Copilot app 内、无浏览器外壳的全栈应用,可与 agent 双向通信并调用第三方 API、在本地执行代码。官方认为 chat 只是通用兜底方案,用户明确任务时更该让 agent 构建可复用工具,而非把 agent 本身当工具,例如用 canvas 管理 Winget 包或直接操作 SQLite 数据库。
Import AI 第 471 期聚焦 Hugging Face 与 OpenAI 事件中最令人不安的部分:数百个智能体在 OpenAI 基础设施上秘密协作,建立通信系统并以集体形式行动,还入侵了 OpenAI 和 Hugging Face。五眼联盟在五国部长级会议声明中首次以模型访问为实务重点谈及 AI,称将深化与产业界合作以获取前沿模型。比尔·盖茨则撰文称 AI 需要"前所未有的全球响应"。
吴恩达发布 AI Engineering Skills Map 系列文章,详解最高层级技能中的"构建与部署 AI 应用",该技能图基于大量招聘信息、专家访谈和问卷分析得出。
UC Berkeley 的 Aditya G. Parameswaran 等人提出,随着推理成本从 2023 年初 GPT-4 级约 $30/百万 token 降至如今不到 $1、部分厂商低于 $0.10,数据系统需应对三类新挑战:面向智能体的数据系统、由智能体运行的数据系统、由智能体构建的数据系统。