Workflow
量子位
icon
Search documents
清华百川楼挂牌启用后,就地圆桌开聊AI医疗
量子位· 2025-12-27 04:59
Core Viewpoint - The discussion emphasizes the importance of not overly aligning AI medical initiatives with traditional medical practices, suggesting that innovation should not be constrained by conventional medical perspectives [1][62]. Group 1: Perspectives on AI in Healthcare - The roundtable featured three key perspectives: AI entrepreneurs, researchers, and healthcare practitioners, highlighting the complexity of integrating AI into the medical field [4][5]. - The future of AI in healthcare is seen as critical, with discussions extending beyond technology to include ethical considerations, decision-making authority, and clinical reasoning [9][10]. Group 2: Vision for AI in Medicine - AI in medicine is viewed as a complex system that reflects the challenges of achieving AGI (Artificial General Intelligence), with medical knowledge spanning multiple disciplines [13][14]. - The development of large medical models is essential, serving as a foundational infrastructure that integrates various types of medical data [16][17]. - AI has the potential to drive advancements in medical research by identifying complex patterns that traditional methods may overlook [19][20]. - The relationship between doctors and patients is expected to evolve, with patients becoming more informed and demanding higher standards from healthcare providers [21][22]. Group 3: AI Medical Benchmarks - The benchmarks for AI in healthcare must evolve to reflect the dynamic nature of AI technology, focusing on long-term health monitoring and adaptive treatment plans [30][31]. - In real medical scenarios, the effectiveness of AI is measured by its clinical reasoning capabilities, acceptance by healthcare professionals, and its impact on treatment outcomes [33][34]. Group 4: Unique Value Proposition of Baichuan Intelligence - Baichuan Intelligence aims to create a companion AI that engages in long-term decision-making rather than providing one-off answers, emphasizing the importance of patient and doctor engagement [37][40]. - The company collaborates with top hospitals while recognizing that professional endorsement does not guarantee product quality [39]. Group 5: Challenges and Recommendations for AI in Healthcare - The regulatory environment in healthcare poses significant challenges for AI innovation, necessitating careful navigation to maintain trust while integrating AI into decision-making processes [50][52]. - Young professionals entering the AI healthcare field are encouraged to find genuine interests and embrace interdisciplinary knowledge to foster innovation [54][56].
一只大头机器狗供不应求,打响了消费级具身智能第一枪
量子位· 2025-12-26 12:28
Core Viewpoint - The article highlights the emergence of Vbot's BoBo, a consumer-grade robotic dog, as a leading product in the embodied intelligence sector, achieving significant sales and consumer interest in a short time frame [5][57][79]. Group 1: Product Performance - Vbot's BoBo sold 1,000 units in just 52 minutes, setting a record in the consumer-grade robotic dog market [7][10]. - The product is priced at ¥9,988, making it accessible while offering high-end specifications, including a computing power of 128 TOPS and a battery capacity of 594Wh, which is 37.5% higher than the industry average [23][34][37]. - BoBo's design incorporates emotional engagement through facial expressions and movements, making it appealing to families, especially children [28][51][56]. Group 2: Market Positioning - Vbot has positioned BoBo as the first brand in the consumer-grade embodied intelligence market, addressing emotional companionship and practical assistance needs [57][75]. - The product's success is attributed to its unique design and technology, which combines advanced AI capabilities with user-friendly interactions, differentiating it from existing robotic products [33][44][46]. Group 3: Technological Innovation - BoBo utilizes a novel VLA (Vision-Language-Action) model and an Agent architecture, allowing it to understand and execute complex tasks based on natural language commands [38][39]. - The integration of a full-scene spatial base model enables BoBo to perform tasks like waking up a child by understanding context and planning routes [32][41]. Group 4: Industry Impact - Vbot's rapid success reflects a shift in the consumer robotics landscape, moving from industrial applications to personal, emotionally engaging products [62][79]. - The article suggests that the acceptance of consumer-grade embodied intelligence products like BoBo could lead to widespread adoption similar to that of smart cars and intelligent driving technologies in the near future [79].
清华唐杰:领域大模型,伪命题
量子位· 2025-12-26 08:52
Group 1 - The core idea is that scaling foundational models through pre-training is essential for AI to acquire world knowledge and basic reasoning capabilities [4][5] - More data, larger parameters, and saturated computation remain the most efficient methods for scaling foundational models [5] - The concept of domain-specific large models is considered a false proposition, as true AGI (Artificial General Intelligence) has not yet been achieved [28][30] Group 2 - Enhancing reasoning capabilities and aligning long-tail abilities are crucial for improving real-world AI performance [6][7] - The introduction of agents marks a significant milestone in AI, allowing models to interact with real environments and generate productivity [10][11] - Implementing memory mechanisms in models is essential for their application in real-world scenarios, with different memory stages mirroring human memory [12][13] Group 3 - Online learning and self-evaluation are key components for models to improve autonomously, with self-assessment being a critical aspect of this process [14][15] - The integration of model development and application is becoming increasingly important, with the goal of replacing human jobs through AI [16][17] - The future of AI applications should focus on enhancing human capabilities rather than merely creating new applications [32][34] Group 4 - Multimodal capabilities are seen as promising, but their contribution to AGI's upper intelligence limit remains uncertain [21][22] - The development of embodied AI faces challenges, including data acquisition and the stability of robotic systems [25][26] - The existence of domain models is driven by enterprises' reluctance to fully embrace AI, aiming to maintain a competitive edge [29][31]
训练时间爆砍80%!港大快手联合打造了一个AI炼金师:专挑“有营养”数据,20%数据达成50%效果
量子位· 2025-12-26 08:52
Alchemist团队 投稿 量子位 | 公众号 QbitAI 想象一下,如果让一个大厨用发霉的食材、过期的调料来做菜,即使厨艺再高超,也做不出美味佳肴。AI训练也是同样的道理。 一、数据就像食材,质量决定成品 现在的AI图像生成模型,如Stable Diffusion、FLUX等,需要从网络上爬取数百万张图片来学习。但这些图片质量参差不齐:有些模糊不 清,有些内容重复,有些甚至只是广告背景图。用这些"食材"训练出来的AI,自然效果不佳。 由香港大学丁凯欣领导,联合华南理工大学周洋以及快手科技Kling团队共同完成的这项研究,开发出了一个名为"炼金师" (Alchemist) 的AI系统。它就像一位挑剔的大厨,能从海量图片数据中精准挑选出最有价值的一半。 更让人惊喜的是: 二、让AI学会"自我评判" 2.1 传统方法的局限 传统的数据筛选方法就像用筛子筛米粒,只能按照单一标准过滤: 这些方法的问题在于: 它们不知道哪些数据真正有助于AI学习 。 2.2 炼金师的智慧 "炼金师"更像是一位经验丰富的美食评委,它能同时考虑多个维度: 用这一半精选数据训练出的模型,竟然比用全部数据训练的表现还要好 训练速度快了 5 ...
AI金矿上打盹的小红书,刚刚醒了一「点点」
量子位· 2025-12-26 08:52
一点进去发现,好家伙,小红书这波操作,终于是 把官方AI整上了我的首页 。 是新功能,但也是老面孔。AI助手名叫 点点 , 用户们应该挺眼熟,就是之前在评论区常会被@的小红书版评论罗伯特。 鱼羊 发自 凹非寺 量子位 | 公众号 QbitAI 事情是这样的。 作为一个小红书重度用户,今天一开软件我天塌了:我的侧边栏呢??? 我赶紧一个搜索,原来官方真是更新了玩法。 评论区@不到了,但现在,你可以在小红书里这样玩AI:笔记直接分享给点点,不用手动跳转,即可开启边刷边聊模式。 还真别说,现在的社交媒体上,要没点AI出没,是有那么点不习惯。 像微博,不止有到处串场的评论罗伯特,也把「智搜」功能插进了每一个热门话题里,主打一个让用户吃瓜不迷路;而微信,也把元宝总结的 功能内置进了公众号文章页面。 看上去在AI上一直比较保守的小红书,现在也醒了一「点点」。 AI一点点,体验变好了吗? AI一点点,有没有让刷社媒的体验变好,还是得实测一波才知底细。 交互体验 先来看看交互方式。 第一种方式,就是在原来首页侧边栏的位置, 点击小气泡进入点点对话框 : 用法跟别的AI助手没有什么不同,好处就是无需跳转其他App,在小红书本书 ...
英伟达成美国大模型开源标杆:Nemotron 3连训练配方都公开,10万亿token数据全放出
量子位· 2025-12-26 06:35
而且开放得很彻底: 不仅开放模型权重,还要把超过10万亿token的训练数据、预训练和后训练软件、训练配方全部公开。 梦晨 发自 凹非寺 量子位 | 公众号 QbitAI 英伟达在开源模型上玩的很激进: "最高效的开放模型家族"Nemotron 3,混合Mamba-Transformer MoE架构、NVFP4低精度训练全用上。 与其他开源模型相比性能有竞争力,且速度快1.5-3.3倍。 把Mamba和Transformer混着用 Nemotron 3在架构层面追求推理效率的最大化。 传统Transformer的自注意力机制需要对不断增长的KV Cache做线性扫描,序列越长,计算开销越大。 英伟达的解决方案是大量使用Mamba-2层替代自注意力层——Mamba层在生成时只需要存储固定大小的状态,不受序列长度影响。 以Nano型号为例,整个模型主要由交替堆叠的Mamba-2层和MoE层构成,自注意力层只保留了少数几个。 论文给出的层排布模式是:5个Mamba-2+MoE的重复单元,接3个同样结构的单元,再来1个包含注意力层的单元,最后是4个Mamba- 2+MoE单元。 在8k输入、16k输出的典型推理场景下 ...
第一批拿12.8万月薪的实习生已经出现!AI人才抢夺战真的好激烈
量子位· 2025-12-26 06:35
衡宇 发自 凹非寺 量子位 | 公众号 QbitAI 好震惊,好意外,现在一份4–6个月的AI相关实习,月薪已经接近14万人民币了! 而且 这个价格不是个例 —— OpenAI、Anthropic、Meta、Google DeepMind等巨头,都为实习、Fellowship、Residency这类短期岗位,开出足以对标全职研究员的价 格。 Business Insider最新披露的一组数据显示,目前AI相关实习和研究型短期项目的月薪,已经普遍来到7000–18000美元区间,折合人民币约 4.9-12.6万元。 换算成年薪水平,是不是 已经明显超出大多数行业对"实习生"这一角色的传统认知 …… 真·AI人才的生活,我的梦 (没错已经开始白日做梦了) 。 书归正传。 继大厂、巨头为成熟的AI人才大动干戈,甚至扎克伯格为了挖OpenAI的人亲自洗手作羹汤端到想挖的人嘴边过后, 这场纷争终于开始波及那 些还没有正式毕业、甚至刚刚进入研究路径不久的人。 水涨船高的AI实习工资 在薪酬层面,实习生、学生研究员、驻留项目,已经可以和全职研究岗站在同一水平线上。 我们先展开来看看硅谷那边的具体情况。 OpenAI Ope ...
超越GPT-5、Gemini Deep Research!人大高瓴AI金融分析师,查数据、画图表、写研报样样精通
量子位· 2025-12-26 06:35
能自动查数据、写分析、画专业金融图表的AI金融分析师来了! 最近,中国人民大学高瓴人工智能学院提出了一个面向真实金融投研场景的多模态研报生成系统—— 玉兰·融观 (Yulan-FinSight) 。 面对用户的研究需求,FinSight能够自动拆解任务,从互联网和金融数据库中搜集包括股价、财报、新闻在内的 多源异构数据 ,并生成包 含"发展历程"、"核心业务架构"、"竞争格局"等章节的 万字图文报告 。 FinSight团队 投稿 量子位 | 公众号 QbitAI △ 可在FinSight预设基础上自行配置 该系统也在 AFAC 2025 金融智能创新大赛挑战组 的1289支队伍中夺冠,并在多项评测中超越了GPT-5 w/Search、OpenAI Deep Research与Gemini-2.5-Pro Deep Research,展现出接近人类专家的金融分析与写作能力。 下面来看详细内容。 为什么通用AI做不好金融研报? 在研究者看来,问题的关键并不在于模型"不会写字",而在于金融行业的研究报告本身是一项 高度结构化、强逻辑、强可视化 的专家级工 作,涉及多个流程。 相比通用问答、检索或文本生成任务,金融 ...
量子位编辑作者招聘
量子位· 2025-12-26 04:24
目前,我们有 三大方向 岗位招聘,希望你是 (或者能成为) 这三个方向的内容专家: 岗位均为全职,工作地点:北京中关村。 我们是一家以 追踪AI新进展 为核心的内容平台,经过8年积累,目前拥有顶流影响力,广泛且备受认可的产业资源,以及时代风口的最佳观 测和学习生态位。 编辑部 发自 凹非寺 量子位 | 公众号 QbitAI AI热潮还在汹涌,但如果你还不知道如何参与……那为什么不来 量子位 呢? AI产业方向 :关注基建层创新,包含芯片、AI Infra、云计算; AI财经方向 :关注AI领域创投和财报,跟踪产业链资本动向; AI产品方向 :关注AI在应用和硬件终端方向的进展。 社招:覆盖编辑、主笔、主编各个层级,按能力匹配岗位; 校招:应届毕业生,接受实习且可转正。 站在AI浪潮之巅 :第一时间接触和了解AI领域最新技术和产品,构建完整的AI认知体系。 玩转AI新工具 :将各种AI新技术、新工具应用于工作,提升工作效率和创造力。 打造个人影响力 :通过撰写独家原创内容,建立个人知名度,成为AI领域的意见领袖。 拓展行业人脉 :与AI领域大咖零距离接触,参与重要科技活动和发布会,拓展行业视野。 获得专业指导 ...
P图新手福音!智能修图Agent一句话精准调用200+专业工具,腾讯混元&厦大出品
量子位· 2025-12-26 04:24
JarvisEvo团队 投稿 量子位 | 公众号 QbitAI 下面就来了解一下详细情况吧~ 自我评估和修正 研究背景与动机 近年来,基于指令的图像编辑模型虽然取得了显著进展,但在追求"专业级"修图体验时,仍面临两大核心挑战: 1. 指令幻觉 (Instruction Hallucination): 现有的文本思维链 (Text-only CoT) 存在信息瓶颈。模型在推理过程中"看不见"中间的修图结果,仅凭文本"脑补"假设进行下一步操作的 视觉结果,容易导致事实性错误,无法确保每一步都符合用户意图。 一句话让照片变大片,比专业软件简单、比AI修图更可控! 腾讯混元携手厦门大学推出 JarvisEvo ——一个统一的图像编辑智能体模拟人类专家设计师,通过 迭代编辑、视觉感知、自我评估和自我反 思 来"p图"。 "像专家一样思考,像工匠一样打磨" 。JarvisEvo不仅能用Lightroom修图,更能"看见"修图后的变化,并自我评判好坏,从而实现无需外部 奖励的自我进化 。 2. 奖励黑客 (Reward Hacking): 在强化学习进行偏好对齐的过程中,策略模型(Policy)是动态更新的,而奖励模型(R ...