00:00 開場:AI 的智慧變便宜以後,最缺的是什麼 01:04 Alexandr Wang(Meta):YC 創業學校 2026 01:59 Wang:第一個點子太早了 02:52 Wang:訓練模型的三樣東西,Scale AI 的起點 04:50 Wang:在共識之前的信念 06:23 Wang:卡住的不是模型,是擴散 07:24 Wang:大衛與歌利亞,小團隊的機會 09:11 Wang:智慧變多以後,稀缺的是願景 11:50 Meta 的處境:Llama 4 之後 12:48 核對「便宜八倍」與資本支出 13:42 對談之後:Muse 系列產品 14:50 Wang:給十八歲自己的建議 15:34 篤定與倖存者偏差:這對一般人的意思 大家好,我是 Jim,這裡是 AI Notebook。這一集是一個新系列,演講訪談系列的第一集,主角是 Alexandr Wang。他十九歲創辦 Scale AI,現在在 Meta 帶領超級智慧實驗室。今年夏天,他回到 YC 的創業學校,對著台下一群十八到二十歲的年輕人,說了一句很大的話。他說,現在可能是一個文明只會遇到一次的機會。這一集只問一個問題,這句話是真的嗎?如果是真的,機會到底在哪裡,跟我們一般人又有什麼關係?我先把答案放在前面。他的意思是,AI 的智慧,正在變成按一下就拿得到的東西。當智慧不再稀少,最缺的,就變成知道要往哪裡走的人。這句話,我覺得有一半站得住,另一半要打折。因為說這句話的人,正好在全世界花最多錢追趕 AI 的公司之一,負責這整件事。我們分三步走。先看他年輕的時候,怎麼看見別人沒看見的東西。再看他說的機會,到底長什麼樣子。最後把他的身分放回來,看看這句話要打幾折。先從 2016 年,一個十九歲的大學生開始。 📎 資料來源 • Alexandr Wang: "This is a Once-in-a-Civilization Opportunity"(YC Startup School 2026, Garry Tan 對談) https://youtu.be/sJ4VJWycX9M • YC 官方逐字稿:Alexandr Wang — Building a Frontier Lab https://www.ycrootaccess.com/p/alexandr-wang-building-a-frontier • Y Combinator:Scale AI 公司頁(S16, Jared Friedman) https://www.ycombinator.com/companies/scale-ai • NYT:Meta Invests $14.3 Billion in Scale AI to Kick-Start Superintelligence Lab(2025-06-13) https://www.nytimes.com/2025/06/12/technology/meta-scale-ai.html • CNBC:Scale AI promotes Droege to CEO as founder Wang heads for Meta(2025-06-12) https://www.cnbc.com/2025/06/12/scale-ai-promotes-strategy-chief-droege-to-ceo-as-wang-heads-for-meta.html • Meta:Personal Superintelligence(Mark Zuckerberg) https://www.meta.com/superintelligence/ • Meta AI Blog:Introducing Muse Spark(2026-04-08) https://ai.meta.com/blog/introducing-muse-spark-msl/ • Meta Newsroom:Introducing Muse Spark https://about.fb.com/news/2026/04/introducing-muse-spark-meta-superintelligence-labs/ • Bloomberg Tech 2026:Meta AI Chief Wang on Winning the Race in AI(2026-06-05) https://youtu.be/bO06SaZdnXs • Meta for Developers:Build with Muse Spark(Muse Spark 1.1 定價) https://dev.meta.ai/resources/blog/build-with-muse-spark/ • Claude API 定價(含舊版 Opus) https://claude.com/pricing • CNBC:Meta Q1 2026 earnings(資本支出指引、裁員) https://www.cnbc.com/2026/04/29/meta-q1-earnings-report-2026.html • CNBC:Meta Q2 2026 earnings(資本支出 1300–1450 億美元) https://www.cnbc.com/2026/07/29/meta-q2-earnings-report-2026.html • Meta for Developers:Meet Muse Spark 1.2 and Muse Code(2026-08-05) https://dev.meta.ai/resources/blog/build-with-muse-code/ • Meta AI Research:Introducing Muse Glimmer(2026-08-10) https://research.meta.ai/blog/introducing-muse-glimmer-open-agentic-model • Axios:Meta debuts Muse, its long-planned personal AI agent(2026-09-08) https://www.axios.com/2026/09/08/meta-debuts-muse-personal-ai-agent • TechCrunch:Everything new coming to Meta's AI agent Muse(2026-09-23) https://techcrunch.com/2026/09/23/everything-new-coming-to-metas-ai-agent-muse/ • 404 Media:Meta Tests Muse AI Agent Calls That Are Actually Made By Humans in a Call Center(2026-09-22) https://www.404media.co/meta-tests-muse-ai-agent-calls-that-are-actually-made-by-humans-in-a-call-center/ • GeekWire:Amazon blocks Meta's Muse AI assistant(2026-09-21) https://www.geekwire.com/2026/amazon-blocks-metas-muse-ai-assistant-in-new-standoff-over-agentic-shopping/ • Le et al., Building High-level Features Using Large Scale Unsupervised Learning(Google, 2012) https://arxiv.org/abs/1112.6209 --- Jim AI Notebook | Note the Future 每日深度解析 AI 最新進展 #AlexandrWang #Meta #MetaSuperintelligenceLabs #ScaleAI #YCombinator #StartupSchool #MuseSpark #AI代理 #創業 #超級智慧 #演講訪談系列 #JimAINotebook
Transcript is not available for this video.