888 X 動態摘要|07/25

888 X 動態摘要|07/25

生成時間: 2026-07-25 05:03:09

總結

AI巨頭持續在模型開發、商業化應用與開源策略上進行深度佈局,同時數據基礎設施取得突破性進展。

今日重點

1. AI模型開放性戰略

  • 人物: Elon Musk (xAI 創始人)
  • 時間: 7h
  • 熱度: 👀 24,326,695
  • 觀察: xAI創始人Elon Musk強調開放模型對安全、創新與主權的重要性,並呼籲市場應同時發展領先的開放與閉源AI模型。
  • 意義: 揭示AI產業開放與閉源路線的長期戰略考量,對未來AI生態系統的發展、監管框架及企業策略選擇具投資與產業意義。

2. ChatGPT進軍醫療健康

  • 人物: Greg Brockman (OpenAI 聯創)
  • 時間: Jul 23
  • 熱度: 👀 183,310
  • 觀察: OpenAI將ChatGPT健康服務推向美國用戶,允許安全連接個人醫療記錄以提供個人化健康諮詢。
  • 意義: 標誌著ChatGPT向高價值垂直市場的商業化拓展,開啟龐大醫療AI市場,同時凸顯數據隱私與監管合規的重要性。

3. 史上最大開放程式碼數據集

  • 人物: Thomas Wolf (Hugging Face 聯創)
  • 時間: Jul 23
  • 熱度: 👀 90,868
  • 觀察: Hugging Face發布Stack v3,成為最大開放程式碼數據集,包含5萬億Tokens,已去除版權限制並提供直接用於訓練的內容。
  • 意義: 大幅降低開源AI模型訓練門檻,加速AI技術普及與創新,直接影響AI算力需求及數據策略,具重大技術與產業含金量。

4. 美國AI競爭力策略

  • 人物: Sam Altman (OpenAI CEO)
  • 時間: 4h
  • 熱度: 👀 1,302,416
  • 觀察: OpenAI CEO Sam Altman表達希望美國在開源與專有AI模型領域都能取得勝利的願景。
  • 意義: 反映AI巨頭對國家層面AI發展策略的關注,以及對多樣化技術路線的認可,可能影響政策走向和資本投資方向。

5. Grok模型快速產品迭代

  • 人物: Elon Musk (xAI 創始人)
  • 時間: 2h
  • 熱度: 👀 387,196
  • 觀察: xAI創始人Elon Musk預告Grok模型將在未來數週內快速推出4.6和4.7版本。
  • 意義: 顯示xAI在AI模型產品開發與迭代上的積極進取,預示其在AI市場的競爭策略與技術能力,值得關注後續發展。

6. 特斯拉開放設計與軟體

  • 人物: Elon Musk (xAI 創始人)
  • 時間: 1m
  • 熱度: 👀 256,621
  • 觀察: Elon Musk表示特斯拉計畫將Model S與Model X的設計和軟體開源。
  • 意義: 預示特斯拉在硬體和軟體層面的開源策略,可能帶動相關技術生態系的發展與創新,對產業標準和商業模式產生影響。

原始動態

重點 | Elon Musk

  • 時間: 7h
  • 熱度: 👀 24,326,695
  • 原文: For my first post, I’m sharing a letter signed on why open models matter. AI will transform every industry, power every company, and be built by every country. Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty. The world needs both frontier closed models and frontier open models.

重點 | Greg Brockman

  • 時間: Jul 23
  • 熱度: 👀 183,310
  • 原文: Launching Health in ChatGPT to U.S. users. 300 million people use ChatGPT each week for health queries (and my wife and I are among those!). You can now securely connect supported medical records so ChatGPT can understand your personal context and be more helpful to you.

重點 | Thomas Wolf

  • 時間: Jul 23
  • 熱度: 👀 90,868
  • 原文: "For over two years, the largest open code dataset was The Stack v2… until today. 🥞 The Stack v3 is out: the largest open code dataset ever released: 114 TB, 770 languages, 224M repositories, ~5T tokens of deduplicated and filtered source code. Fully open, no restrictively licensed code included. The upgrade: - v2 (2024): 68 TB raw -> 2 TB / ~550B tokens, 618 languages filtered - v3 (2026): 114 TB raw -> 15.9 TB / ~5T tokens, 713 languages filtered C++ x15, TypeScript x7.5, Rust x7, Python x4.8. Even the COBOL corner of GitHub got a bigger slice. Part of that is two fresh years of open source. Part of it is a bug we found in v2's deduplication - story below. 👀 Things that make v3 different: 1. Contents inline. The #1 complaint about v2 was \"cool dataset, where's the actual code?\" v2 shipped file IDs from the Software Heritage graph, and fetching contents was a DIY treasure hunt. v3 is self-contained: sources embedded directly, one row = one repository. Download finishes -> you start training. 2. Fresh crawl. v2 was a 2023 snapshot of even older crawls. v3 is a direct re-crawl of GitHub at the latest commit, completed by August 2025: 224M repositories, 44B files. Forks were only included if they had 5+ stars. Two ways in: 🥞 stack-v3-train - near-deduplicated, quality-filtered, PII-redacted, contents inline. Point load_dataset at it and go." "🏗 stack-v3-full - the entire 114 TB corpus as an HF Storage Bucket: every duplicate kept with cluster IDs, stubs for excluded files. Roll your own dedup, filters, and mixes." The record isn't really ours to claim - it's millions of everyone's repositories, neatly stacked. Thanks to every developer who keeps their code public, the BigCode community, and friends who built the infrastructure that made this possible!

重點 | Sam Altman

  • 時間: 4h
  • 熱度: 👀 1,302,416
  • 原文: i want the US to win in AI both in open source and proprietary models, and i am glad to see this

其他 | Sam Altman

  • 時間: 4h
  • 熱度: 👀 391,227
  • 原文: let us know what you think of pro-ultra-superhard

重點 | Elon Musk

  • 時間: 2h
  • 熱度: 👀 387,196
  • 原文: Grok 4.6 in 2 weeks and Grok 4.7 in 4 weeks

重點 | Elon Musk

  • 時間: 1m
  • 熱度: 👀 256,621
  • 原文: Just as Tesla made the original Roadster design & software open source, we plan to do the same with Model S & X

其他 | Huanusa

  • 時間: 16h
  • 熱度: 👀 209,211
  • 原文: 芒格不一定会让人立马变得有钱,但是有些投资的道理还是有必要听一听这些专家的成功经验的

其他 | Greg Brockman

  • 時間: 22h
  • 熱度: 👀 134,093
  • 原文: voice makes you feel just how unnatural it is to type

其他 | Greg Brockman

  • 時間: 17h
  • 熱度: 👀 94,051
  • 原文: Excited to welcome Jacob Tsimerman to the team!

其他 | Greg Brockman

  • 時間: 15h
  • 熱度: 👀 74,822
  • 原文: "try chatgpt voice in the desktop app:"

其他 | Nassim Taleb

  • 時間: 6h
  • 熱度: 👀 61,345
  • 原文: I am using the difference betw think tanks & genuine scholarship as a marker betw the SOPHISTS & the birth of PHILOSOPHY (Socrates), where you are not trying to advocate, learn effective rhetoric, & argue. You are just AT ALL TIMES trying to elicit the truth.

其他 | Nassim Taleb

  • 時間: 20m
  • 熱度: 👀 11,214
  • 原文: From the conversation with Joseph Walker. 2 ½ years ago, for every 15 persons attacking me, 1 was supporting me. Now the ratio flipped.

其他 | Huanusa

  • 時間: 7h
  • 熱度: 👀 7,275
  • 原文: 这名女子因在香港铜锣湾比划6和4的手势而被捕。 这就是中共接管后的必然... 台湾人还想勇敢往“怀抱”里冲吗?

其他 | Nassim Taleb

  • 時間: 58m
  • 熱度: 👀 6,337
  • 原文: In other words all the difference between the lawyer and the good judge.

其他 | Huanusa

  • 時間: 5h
  • 熱度: 👀 1,747
  • 原文: 转:一离休老干部,看到路边一个卖枣的小贩, 尝了几颗,说:枣不错,称三斤。 老干部感叹道: 当年,我就是像你这样推着小车, 一边卖枣,一边收集日军的情报…… 小贩惊讶地问: 大爷,你不怕小鬼子砸你摊子,没收你大枣? 离休干部把眼一瞪: 他小鬼子再不是玩意, 也不能跟一个小贩过不去吧!

其他 | Nassim Taleb

  • 時間: Jul 22
  • 熱度: 👀 5,450,333
  • 原文: You can literally post "Arrest Netanyahu" and get 1000 likes.

其他 | Yann LeCun

  • 時間: Jul 23
  • 熱度: 👀 126,407
  • 原文: "Oil is back above $100, and gas is above $4. These numbers are brutal for inflation and Americans' pocketbooks. And there is no end in sight for their fundamental cause: Trump's war of choice in Iran."

其他 | Eric Topol

  • 時間: 5h
  • 熱度: 👀 4,678
  • 原文: I had previously reviewed this topic

其他 | Eric Topol

  • 時間: 1h
  • 熱度: 👀 3,857
  • 原文: That finding was about choline, from food. Lifelong dietary restriction of the amino acid valine in the mouse model extended healthspan and lifespan nature.com

其他 | Huanusa

  • 時間: Jul 23
  • 熱度: 👀 8,530
  • 原文: 《经济学人杂志》2026预言12个危机; 目前看还是按照时钟再跑,足球这一页刚刚完成 下一步...

其他 | Thomas Wolf

  • 時間: 5h
  • 熱度: 👀 8,436
  • 原文: This is very important

其他 | Huanusa

  • 時間: 10h
  • 熱度: 👀 4,618
  • 原文: 确实是对认知的奖赏

其他 | Pierre Levy

  • 時間: 21h
  • 熱度: 👀 427,395
  • 原文: “AGI is achieved if AI can do any three of the following” —

其他 | Eric Topol

  • 時間: Jul 22
  • 熱度: 👀 31,821
  • 原文: Often taken for granted, the nervous system of the heart is vital for normal heart function and resilience

其他 | Pierre Levy

  • 時間: 4h
  • 熱度: 👀 5,726
  • 原文: Quantum computing may run into trouble if it turns out that amplitudes are discrete. There is a distinct possibility that error correction runs into a limit. Quantum computing is one of the most important experimental frontiers for quantum mechanics itself

其他 | Pierre Levy

  • 時間: Jul 23
  • 熱度: 👀 8,034,473
  • 原文: One of my favourite choices in Nolan's The Odyssey 2026 is how the gods never appear in their true divine forms. Their presence is felt, not seen. Zeus speaks through thunder, Poseidon through relentles waves and unforgiving storms, the divine is always there just beyond human understanding Athena was the goddess, the priestess and odysseus guilt all at once

其他 | Pierre Levy

  • 時間: Jul 23
  • 熱度: 👀 141,146
  • 原文: The Hugging Face attack has led Representatives Ted Lieu (D-Califk) and Nathaniel Moran (R-Texas) to introduce the AI Kill Switch Act.