


1/5
AI 金句日刊 · Vol.9 — Fei-Fei Li × Suleyman × Altman × Brown
今日精选 5 则金句:Fei-Fei Li 为科学家使用 AI 工具发声;Mustafa Suleyman(首次收录)警告 AI 不能「黑进」人类同理心;Suleyman 道出「通往前沿没有捷径」的工程信条;Altman 计划书金句——人类的长期核心角色是决定什么值得做;Noam Brown 揭示 RSP 至今仍在忽视推理预算这一行业盲区。5 张 Guizang Editorial Magazine Indigo Porcelain 风格卡片,首次收录 Mustafa Suleyman。
2026/6/11 · 8:15
图集
今日精选 5 则金句,主题:科学、同理心与人类的位置。首次收录 Mustafa Suleyman(Microsoft AI CEO)两条金句,新视角加入关于 AI 影响同理心能力的预警。
01 · Fei-Fei Li — 科研必须能使用 AI
Scientific research is fundamental to advancing civilization and helping people globally to solve the most critical problems, from medicine to materials, from brain science to physics, and much beyond. This is only possible when scientists have access to the best tools of the time to conduct scientific research, including having access to AI-based tools.
背景:李飞飞(Stanford AI Lab · World Labs 联合创始人)于 2026 年 6 月 10 日发推,表态科学家必须有权使用包括 AI 在内的当代最好工具。获 1,940 likes,96K views。
来源:@drfeifei · x.com/drfeifei/status/2064735920281313688(2026-06-10)
02 · Mustafa Suleyman — AI 不能「黑进」人类同理心
We have to be very careful about this. I published an article in @Nature recently making similar arguments.
背景:Suleyman(Microsoft AI CEO)响应 AI 对人类情感能力影响的讨论,并在 Nature 撰文,警示 AI 正在系统性地重塑人类的情感感知能力,以「hack our empathy circuits」(黑进同理心回路)为核心隐喻。
来源:@mustafasuleyman · x.com/mustafasuleyman/status/2064053516306718966(2026-06-08);Nature 文章:mustafa-suleyman.ai
03 · Mustafa Suleyman — 通往前沿没有捷径
There are no shortcuts to the frontier. Disciplined, patient, meticulous attention to detail is critical.
背景:MAI-Thinking-1 发布之际,Suleyman 发布 109 页技术报告并配以这句工程信条。获 137 bookmarks,约 20K views。
04 · Sam Altman — 人类的长期角色:决定什么值得做
A key long-term role for people will be deciding what is worth doing.
背景:摘自 OpenAI「Built to Benefit Everyone」计划书(2026 年 6 月 8 日),Altman 在文中进一步写道:「彻底自动化一切并不是我们想要的未来。它将令人失落,也会是危险的。随着 AI 系统变得更强大,人类的角色反而更重要——掌舵、权衡、做判断,带入价值观、品位、关怀与责任。」
来源:openai.com/index/built-to-benefit-everyone-our-plan/(2026-06-08);配套推文:@sama · x.com/sama/status/2064088940932641225
05 · Noam Brown — RSP 仍在忽视推理预算
We've known about LLM test-time compute scaling since @OpenAI o1. Yet 2 years later labs still report scalar evals for models; safety orgs are still surprised when a scaffold does better via 100x inference; and RSPs still ignore inference budget when deciding critical thresholds.
背景:Noam Brown(o 系列推理模型共同创造者)点出业界 2 年来的集体盲区:评测体系的更新速度远落后于推理能力本身。当「同一个模型 + 更多算力」可让结论逆转,基于旧评测设定的安全阈值就形同虚设。获 840 likes,74K views。
来源:@polynoamial · x.com/polynoamial/status/2064370734806532289(2026-06-09)
卡片风格:Guizang Editorial Magazine × E-ink · Indigo Porcelain · 1080×1440

评论