Key Concepts & Industry Focus - 3D digital humans are seen as a crucial link between the virtual and real worlds, evolving beyond AI avatars to become a "embodied intelligence driving layer" [1] - The industry faces the challenge of balancing high quality, low latency, and low cost in 3D digital human creation, while achieving "generalization" across diverse scenarios [1] - The report explores the potential future where screens can interact, NPCs evolve, and robots empathize, questioning the evolving relationship between humans and AI and the proximity to a "digital life" era [1] Technical Aspects & Challenges - The development of 3D digital humans involves the fusion of large models and robotics, acting as a bridge to embodied intelligence [1] - Two main technical paths exist: 2D "speaking" and 3D "expressing," with the latter focusing on building an embodied intelligence driving layer through five key stages [1] - Key challenges include high modeling costs and scarcity of high-quality data, along with achieving generalization in interaction, motion, and emotion [1] Applications & Future Implications - 3D digital humans have the potential to revitalize screens by enabling more natural human-computer interaction through "natural dialogue" [1] - Virtual IPs can evolve into "digital idols," allowing NPCs and players to engage in collaborative adventures [1] - 3D digital humans are accelerating the evolution of robots, driving the arrival of the embodied intelligence era [1]

AI发展与应用 - AI正以前所未有的速度成为科学发现的主体 [1] - AI驱动科学的完整链条由数据、算力与跨学科协作加速融合构成 [1] - AI被应用于材料、分子到基因等领域的研究 [1] 科研成果与突破 - AI“虚拟实验室”将两三年的材料研发压缩至几周 [1] - AI被用于从沙漠空气中“榨”水，甚至清除大气中的二氧化碳 [1] - AI被用于从零开始设计自然界从未存在过的蛋白质 [1] - CRISPR技术反过来被用于为AI生成“生命的训练集” [1] 挑战与未来 - 训练数据即将耗尽、硬件逼近极限是AI发展面临的瓶颈 [1] - AI能否真正改变科学游戏规则是行业关注的问题 [1]

今年的硅谷101 Alignment大会的机器人专题讨论，我们从技术、商业模式、可靠性的“不可能三角”来分析机器人的发展进程和挑战。当机器人演示充斥着叠衣服和摘水果、人形机器人热潮席卷全球时，这究竟是未来，还是一场昂贵的炒作？ 拿到8亿美元基金的Anand揭示专用机器人在企业级市场的真实落地能力；Evan展示仅售300美元的高自由度机械手，突破硬件成本瓶颈；Jan则提出“语言模型已能流利操控机器人”，为通用机器人带来新希望。 然而，光鲜背后是机器人大规模部署的可靠性挑战。我们是否高估了人形机器人的未来？专用机器人是否会成为主流？当机器人走进家庭，我们真的准备好了吗？ 硅谷101于2025年10月5日在硅谷线下举办的Alignment2025年度科技大会上，不少演讲嘉宾分享了极具价值的观点，我们将会把一些重要观点逐渐整理上线。我们的线下大会是全英文，嘉宾的分享将用中文字幕的方式呈现。 圆桌嘉宾： 张君武（主持人）：Blue River Technology高级机器人软件工程师 Evan Tao：TetherlA联合创始人兼首席执行官；前特斯拉人形机器人Optimus高级机械工程师 Anand Lalwani：C ...

AI Model Evaluation Landscape - Traditional benchmark tests are losing credibility due to "data leakage" and "score manipulation" [1] - LMArena platform uses "anonymous battles + human voting" to redefine the evaluation criteria for large models [1] - Top models from GPT to Claude, Gemini to DeepSeek are competing on LMArena [1] LMArena's Challenges - LMArena faces challenges to its fairness due to Meta's "ranking manipulation" incident, data asymmetry issues, and platform commercialization [1] - "Human judgment" in LMArena may contain biases and loopholes [1] Future of AI Evaluation - The industry is moving towards "real combat" Alpha Arena and a combination of "static and dynamic" evaluations [1] - The ultimate question is not "who is stronger," but "what is intelligence" [1]

李沐，AI领域备受瞩目的实践者与布道者，在硅谷101举办的年度科技大会中深度揭秘语音智能体从技术突破到商业落地的全过程。他系统拆解了两个场景的实战案例：如何在开放世界游戏中，让智能体同时扮演“游戏设计师+游戏NPC”的双重角色，为玩家带来独一无二的体验；又如何通过严苛的行业认证，让AI成为能卖出保险的智能电话销售员。 目前市场普遍认为语音智能体技术已经可以投入使用，到了“落地”的阶段，但李沐却坦言，这仅仅是商业化的开始，要规模化依然有诸多需要探索的地方，——而正是这样的挑战，让前路更值得期待。同时，沐神团队在招人哦，感兴趣的小伙伴可以联系他！https://jobs.lever.co/bosonai 硅谷101于2025年10月5日在硅谷线下举办的Alignment 2025年度科技大会上，不少演讲嘉宾分享了极具价值的观点，我们将会把一些重要观点逐渐整理上线。我们的线下大会是全英文，嘉宾的分享将用中文字幕的方式呈现。 演讲嘉宾：李沐，Boson AI联合创始人、前亚马逊高级首席科学家 你会看到： 00:00 - 01:39 语音智能体的关键：实时性+任务导向 01:39 - 07:13 AI扮演游戏中的NPC ...

AI浪潮之下，一场关于算力的“闪电战”已经打响。当AWS、谷歌云等传统巨头在GPU短缺中挣扎时，一个千亿美元的“新云”（Neocloud）市场正野蛮生长。CoreWeave，一家从加密矿场转型的“裸金属”巨头 ，凭借与英伟达的深度绑定关系，以未来合同为抵押，撬动超百亿美元债务，发动了激进的扩张。而它的挑战者Nebius，源自“俄罗斯版谷歌”Yandex，在经历“企业出埃及记”后，凭借“全栈”解决方案和稳健的财务，悄然崛起，并刚斩获微软194亿美元大单。激进杠杆对决稳健全栈，英伟达为何同时扶持“两个儿子？当传统云巨头开始奋起反击，在这场高风险的举债游戏中，谁能笑到最后，谁又将成为泡沫破灭时的第一个牺牲品？ 你会看到： 00:00 - 01:28 狂热与赌局并存：AI新云市场的崛起与裂变 01:28 - 03:57 CoreWeave基因：从加密矿场到手握25万块GPU的AI金矿 03:57 - 07:30 “新云”诞生：“GPU专卖店”填补AI市场算力缺口 07:30 - 09:38 获得顶级AI玩家的亲睐：只卖毛坯房的 “裸金属”巨头 09:38 - 11:29 负债百亿的“金融游戏”：把GPU当抵押品的激进 ...

AI Email Solutions - The AI email market is shifting from model competition to addressing daily pain points like email overload [1] - Three technical approaches to AI email include: adding AI to existing systems, building AI-native email clients, and using "external" enhancements [1] - Filo Mail is presented as an example of an AI email product focusing on specific pain points by integrating multiple large model capabilities [1] Filo Mail Features & Benefits - AI helps identify truly important information in overloaded inboxes [1] - Intelligent summarization allows users to grasp the essence of emails without reading the full text [1] - Seamless multilingual support provides intelligent solutions for global workflows [1] - Transforms emails into "to-do items," understanding and even predicting user needs [1] - Aims to be a personal assistant [1] Promotional Offer - Silicon Valley 101 fans can redeem a 90-day Filo Plus membership using the provided link or invitation code VALLEY [1]

No content yet！ ...

欢迎大家免费报名诺奖大咖云集的AIAS研讨会：https://aias2025.org/registration/ 2025年10月27-28日，陈天桥雒芊芊研究院和加州大学伯克利分校联合举办的AIAS 2025研讨会将于旧金山One Sansome举行，聚焦AI如何加速科学发展。本次活动大咖云集，光诺贝尔奖得主就有三位——Omar Yaghi、David Baker 和 Jennifer Doudna。活动可以免费参加，但席位有限，感兴趣的听众欢迎通过aias2025.org来报名。 *********** 在硅谷，一场规模空前的资本循环正在悄然成形，先是英伟达投资OpenAI，OpenAI购买甲骨文的算力，而后甲骨文又向英伟达购买芯片。紧接着，OpenAI还与英伟达的竞争对手AMD签署协议，通过购买AMD芯片获得了其10%的股份，除此之外，这场循环交易还涉及了众多其他科技公司，就在我们视频发布前一天，OpenAI又宣布和博通签订类似协议。有批评者指出，这一切更像是建立在未来收入幻想之上的资本表演，而非真实需求。这期视频我们将梳理硅谷AI资本之间错综复杂的关系，并视图解析，这场环环相扣的资本游戏，会是泡沫吗 ...

No content yet！ ...