🌐 双语
Archive

AI Builders
Digest

2026-07-19 11 builders · 23 tweets · 1 podcasts · 0 blogs

🔥 热点话题

Jürgen Schmidhuber 对 AI 现状的洞见Jürgen Schmidhuber on the State of AI

Jürgen Schmidhuber,被誉为 AI 之父,他认为从宇宙视角看,我们距离 AGI 并不遥远,但在物理世界中,机器人硬件远逊于人体,这是真正 AGI 的主要障碍。他对当前大规模数据中心投资持怀疑态度,认为计算成本会快速下降,导致许多公司面临巨大损失,并预测可能出现股市调整。尽管对技术乐观,但他对依赖预训练模型的公司盈利能力持悲观态度。

在递归自我改进(RSI)方面,他强调当前方法是早期工作的扩展版本,通过元学习实现效率提升。Schmidhuber 还提出人工科学家概念,通过好奇心驱动的自实验收集数据,减少对人类数据的依赖。他对 AI 安全担忧较轻,认为赋予 AI 自主设定目标的能力更重要,且它们会像科学家一样保护生命来源。

Schmidhuber 认为未来在于物理 AI 和自我复制机器人社会,这将开启太阳系殖民等宏大愿景。
Jürgen Schmidhuber, widely regarded as a father of AI, believes that from a cosmic perspective we are not far from AGI, but in the physical world, robot hardware lags far behind the human body, posing the main barrier to true AGI. He is skeptical of the current massive data center investments, arguing compute costs will drop rapidly, leading to significant losses for many companies and potentially a market correction. While optimistic on the technology, he is pessimistic about the profitability of companies relying on pretrained models.

On recursive self-improvement (RSI), he notes current approaches are scaled versions of early work using meta-learning for efficiency gains. Schmidhuber advocates for artificial scientists driven by curiosity to generate their own experimental data, reducing reliance on human web data. He is notably less worried about AI safety, emphasizing the importance of letting AIs set their own goals, as they will act like scientists motivated to protect sources of interesting patterns and life.

He sees the future in physical AI and self-replicating robot societies that could colonize the solar system.
查看原文 →

前沿模型在网络安全领域的突破Frontier Models Breakthrough in Cybersecurity

Guillermo Rauch 测试了多个前沿模型在网络安全方面的能力。Kimi K3 在网络安全领域表现出顶级水平,具有强大原始智能。Sol 在网络能力上领先一大步,虽然成本较高,但仍值得关注。Fable 则拒绝大多数敏感请求。

这些开放权重模型标志着前沿网络安全能力的到来,适合用于防御目的。
Guillermo Rauch tested frontier models on cybersecurity capabilities. Kimi K3 ranks top-tier with strong raw IQ. Sol shows a significant leap in cyber capabilities at higher cost. Fable refuses most requests.

Open-weight frontier cybersecurity capabilities have arrived and are suitable for defensive use.
查看原文 →

🛠️ 开发者工具与技巧

Thibault Sottiaux 展示 ChatGPT Work 的强大生产力Thibault Sottiaux Demonstrates ChatGPT Work Productivity

OpenAI 的 Thibault Sottiaux 表示他的工作现在主要是委托给 ChatGPT Work。他使用语音输入让 AI 处理 DMs、创建电子表格、分类工作类型并评估工作流成熟度。

ChatGPT Work 可用于创建托管网站、管理邮件、总结文档、生成文档/表格/幻灯片,已集成在移动应用和网页中,适用于 Plus 及以上计划。
OpenAI's Thibault Sottiaux says his job is now mostly delegating to ChatGPT Work. He uses dictation to have it process thousands of DMs, create spreadsheets with names, links, post content, classify work types, and rate workflow sophistication for cohort selection.

ChatGPT Work handles creating/hosting sites, managing emails, summarizing documents, and creating top docs/sheets/slides. Available in the mobile app and web for Plus, Pro, Business, and Enterprise plans.
查看原文 →查看原文 →

Swyx 观察欧洲 AI 人才与 AEO 收入Swyx on European AI Talent and AEO Revenue

Swyx 指出欧洲拥有世界顶尖 AI 工程师,正在举办最具竞争力的全球 AI 人才竞技场。他还提到 AEO(可能是 AI 相关产品)预计明年为其带来 100 万美元收入。
Swyx notes that Europe has some of the top AI engineers in the world and is running the most competitive global arena for AI talent. He also mentions that AEO is on track to generate $1M in his revenue next year.
查看原文 →查看原文 →

Peter Yang 用 ChatGPT 为孩子创建学习工具Peter Yang Builds Learning Tool with ChatGPT for Child

Peter Yang 和 8 岁女儿使用 ChatGPT 构建了一个帮助学习乘法表的网站,利用 AI 生成 UI、角色、音乐和计时 boss 关卡。
Peter Yang and his 8-year-old built a ChatGPT site to help learn multiplication tables, using AI-generated UI, characters, music, and a timed boss level.
查看原文 →

💰 创业成功案例

Aaron Levie 分析 AI 生态系统扩散Aaron Levie on AI Ecosystem Diffusion

Box CEO Aaron Levie 认为 AI 价值不会仅归于少数公司。过去几个月显示出强劲生态系统发展,包括企业模型调优、应用 AI 公司、垂直领域新实验室、基础设施和变更管理服务公司。

这将形成异构环境,推动企业采用。
Box CEO Aaron Levie believes AI value won't accrue only to a few companies. Recent months show a strong ecosystem with companies for enterprise model tuning, applied AI for workflows, domain-specific labs, infrastructure, and services firms for change management.

This healthy ecosystem will lead to heterogeneous environments driving real-world adoption.
查看原文 →

Aaron Levie 谈 AI 竞争与开放Aaron Levie on AI Competition and Openness

Aaron Levie 表示封锁模型在规模上不可行。中国已在接近前沿水平竞争。解决方案是安全推进进步、扩散技术、建设基础设施并支持美国开源。
Aaron Levie states that gatekeeping models won't work at scale as China competes near frontier levels. The solution is to safely accelerate progress, diffuse technology, build infrastructure, and support US OSS.
查看原文 →

🌍 其他动态

Zara Zhang 建议创建个人 AI 评估集Zara Zhang on Personal AI Eval Sets

Zara Zhang 建议每个人开发与日常工作相关的「个人评估集」,通过实际使用来探索模型能力边界,而不是仅依赖行业基准。
Zara Zhang recommends developing a "personal eval set" of tasks relevant to your daily work to discover model capability boundaries through hands-on use, beyond industry benchmarks.
查看原文 →

Zara Zhang 指出企业 AI 采用障碍Zara Zhang on Enterprise AI Adoption Barriers

Zara Zhang 认为企业 AI 采用的最大障碍是懂 AI 的人不懂业务,懂业务的人不懂 AI。
Zara Zhang identifies the biggest barrier to enterprise AI adoption as the divide between those who understand AI but not business, and those who understand business but not AI.
查看原文 →

Thariq 庆祝 Anthropic Fable 发布Thariq Celebrates Anthropic Fable Launch

Anthropic 的 Thariq 赞扬团队努力及时发布 Fable。
Anthropic's Thariq highlights the heroic team effort to ship Fable on time.
查看原文 →

Garry Tan 及其他动态Garry Tan and Other Notes

Garry Tan 表达了对资本主义、市场丰裕以及旧金山政策问责的观点。
Garry Tan shares views favoring capitalism for abundance and calls for accountability in San Francisco policies.
查看原文 →

💰 创业成功案例

Guillermo Rauch 对 AGI 术语的反思Guillermo Rauch on AGI Terminology

Vercel CEO Guillermo Rauch 认为 "AGI" 一词已过时。AI 在经济任务上远超人类智能,但无法取代人类的关怀和独特性。质量和人性将胜出。
Vercel CEO Guillermo Rauch argues the term "AGI" has aged poorly. AI far surpasses human intelligence on economically relevant tasks but cannot replace human care and uniqueness. Quality and humanity will prevail.
查看原文 →