[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-jensen-huang-says-agi-arrived-did-it-zh":3,"article-related-jensen-huang-says-agi-arrived-did-it-zh":29,"series-industry-d5dc9a86-fe85-4066-8ee8-90263c3e6b12":88},{"id":4,"slug":5,"title":6,"content":7,"summary":8,"source":9,"source_url":10,"author":11,"image_url":12,"cover_image":12,"category":13,"language":14,"translated_content":11,"related_article_id":15,"keywords":16,"key_takeaways":11,"views":26,"created_at":27,"published_at":28,"topic_cluster_id":11},"d5dc9a86-fe85-4066-8ee8-90263c3e6b12","jensen-huang-says-agi-arrived-did-it-zh","黃仁勳說 AGI 到了？先別急","\u003Cp>\u003Ca href=\"\u002Fnews\u002Fjensen-huang-says-agi-is-already-here-zh\">黃仁勳說\u003C\u002Fa>，\u003Ca href=\"\u002Fnews\u002Fjensen-huang-agi-definition-deepmind-benchmarks-zh\">AGI\u003C\u002Fa> 可能已經到了。這句話很猛，但他講的不是科幻版。是很務實的版本：AI 代理能做出有用產品、上線、賺到錢，就算跨線。\u003C\u002Fp>\u003Cp>他是 \u003Ca href=\"https:\u002F\u002Fwww.nvidia.com\" target=\"_blank\" rel=\"noopener\">NVIDIA\u003C\u002Fa> 執行長。這家公司供應了大量 AI 晶片。當 \u003Ca href=\"https:\u002F\u002Fopenai.com\" target=\"_blank\" rel=\"noopener\">OpenAI\u003C\u002Fa>、\u003Ca href=\"https:\u002F\u002Fwww.anthropic.com\" target=\"_blank\" rel=\"noopener\">Anthropic\u003C\u002Fa>、\u003Ca href=\"https:\u002F\u002Fdeepmind.google\" target=\"_blank\" rel=\"noopener\">Google DeepMind\u003C\u002Fa> 都在推更會寫程式、會用工具的模型時，這種說法就不是空話。\u003C\u002Fp>\u003Cp>但問題也很明顯。AGI 從來沒有單一標準。黃仁勳講的是商業門檻。很多研究者講的是更廣的能力。也就是說，能不能像人一樣處理大多數智力工作。\u003C\u002Fp>\u003Ch2>黃仁勳到底說了什麼\u003C\u002Fh2>\u003Cp>黃仁勳的說法，出現在 \u003Ca href=\"https:\u002F\u002Fwww.lexfridman.com\u002Fpodcast\" target=\"_blank\" rel=\"noopener\">Lex Fridman Podcast\u003C\u002Fa>。他談的是 agentic AI。也就是 AI 不只會回答，還會規劃步驟、呼叫工具、持續執行任務。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775132875223-ozgf.png\" alt=\"黃仁勳說 AGI 到了？先別急\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>他提到一些開源 agent 工具，像是 \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fopenclaw\" target=\"_blank\" rel=\"noopener\">OpenClaw\u003C\u002Fa>。這類系統可以拆解工作、串接流程，甚至做出能賺錢的產品。他給的標準很直白：如果軟體能自己做出一個爆紅 App，還能每個用戶賺到 0.5 美元，那就算實用版的通用智能。\u003C\u002Fp>\u003Cp>這個標準跟研究界很不一樣。研究者常談的是跨領域泛化。也就是今天學會寫程式，明天也能做數學、法律、醫療推理。黃仁勳談的是另一件事。只要 AI 能在商業環境裡自己做事，很多公司就會說它夠用了。\u003C\u002Fp>\u003Cul>\u003Cli>黃仁勳看的是輸出，有沒有產值。\u003C\u002Fli>\u003Cli>研究界看的是廣度，有沒有跨域能力。\u003C\u002Fli>\u003Cli>Agent 平台把模型接上工具與記憶。\u003C\u002Fli>\u003Cli>收入好算，但不等於真正懂推理。\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>為什麼定義之爭很重要\u003C\u002Fh2>\u003Cp>AGI 爭論常卡住，是因為大家在講不同東西。有人把它當研究目標。有人把它當商業門檻。前者在乎模型能不能像人類一樣學習與適應。後者在乎它能不能真的產出價值。\u003C\u002Fp>\u003Cp>黃仁勳明顯站在後者。他的重點不是哲學。他在意的是，AI 能不能在有限環境內完成多步驟任務，還能交付商業結果。這也很符合 NVIDIA 的市場位置。企業只要能更快做出產品，名詞其實沒那麼重要。\u003C\u002Fp>\u003Cp>但這個差別不能混著看。把今天的模型直接叫 AGI，容易讓人以為難題都解完了。其實沒有。模型還是會幻覺，還是會在長鏈任務裡失焦，還是會在場景一變就整個歪掉。\u003C\u002Fp>\u003Cblockquote>“This is the worst it’s ever going to be.” — Sam Altman\u003C\u002Fblockquote>\u003Cp>這句話常被拿來講 AI 現況。意思很簡單。今天的缺點，可能只是未來的起點。可這不代表現在就是人類等級。它只代表天花板還在往上抬，而且速度不慢。\u003C\u002Fp>\u003Ch2>真正的主角是 agentic AI\u003C\u002Fh2>\u003Cp>黃仁勳談的核心，其實是 agentic AI。這類系統不只是聊天。它會規劃、會呼叫 API、會記憶上下文，還會自己往下做。這跟早期 chat\u003Ca href=\"\u002Fnews\u002Fgoogles-turboquant-cut-llm-memory-sixfold-zh\">bo\u003C\u002Fa>t 很不一樣。早期模型多半只等下一句 prompt。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775132881397-6i31.png\" alt=\"黃仁勳說 AGI 到了？先別急\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>這也是企業願意買單的地方。AI 如果能進工作流，就比較容易算 ROI。它可以幫忙開 ticket、整理會議紀錄、搜尋內部資料、寫程式，還能觸發其他軟體動作。講白了，就是開始像數位員工。\u003C\u002Fp>\u003Cp>開源社群也推了這波。很多團隊不想只做文字生成。他們要的是能執行任務的系統。這讓 agent orchestration 變成熱門題目。模型不再只是回答問題，而是要接住後面的工作。\u003C\u002Fp>\u003Cul>\u003Cli>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fopenai\u002Fopenai-cookbook\" target=\"_blank\" rel=\"noopener\">OpenAI Cookbook\u003C\u002Fa> 已把 tool use 當成基本功。\u003C\u002Fli>\u003Cli>\u003Ca href=\"https:\u002F\u002Fdocs.anthropic.com\u002Fen\u002Fdocs\u002Fagents-and-tools\" target=\"_blank\" rel=\"noopener\">Anthropic 的 agents 文件\u003C\u002Fa> 直接示範工具整合。\u003C\u002Fli>\u003Cli>\u003Ca href=\"https:\u002F\u002Fdeepmind.google\u002Ftechnologies\u002Fgemini\u002F\" target=\"_blank\" rel=\"noopener\">Gemini\u003C\u002Fa> 主打多模態與工具能力。\u003C\u002Fli>\u003Cli>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Flangchain-ai\u002Flangchain\" target=\"_blank\" rel=\"noopener\">LangChain\u003C\u002Fa> 會紅，就是因為大家要工作流，不只是文字。\u003C\u002Fli>\u003C\u002Ful>\u003Cp>所以黃仁勳這番話，比起哲學宣言，更像市場訊號。AI 正在從「會不會聊天」，轉成「能不能做事」。一旦走到這裡，智能的定義就開始像產品規格。\u003C\u002Fp>\u003Ch2>數據和競品怎麼看\u003C\u002Fh2>\u003Cp>如果把黃仁勳的說法，拿去對照現在的模型能力，落差還在。今天最強的模型，確實能寫程式、整理文件、做摘要、跑 benchmark。這些表現很亮眼。可是它們也會在簡單任務上翻車。\u003C\u002Fp>\u003Cp>問題不只在準確率。AGI 如果要成立，應該要穩定。人類助理可以記得例外條件，也能從錯誤裡修正。多數 AI 系統還做不到。它們常需要 guardrail、人工審核、重試機制。\u003C\u002Fp>\u003Cp>把幾個主流方向放一起看，差異就很清楚：\u003C\u002Fp>\u003Cul>\u003Cli>前沿模型在 coding benchmark 常拿高分，但不等於真實工作穩定。\u003C\u002Fli>\u003Cli>Agent 系統能跑多步驟流程，但長任務還是常失控。\u003C\u002Fli>\u003Cli>人會犯錯，但通常懂上下文，也知道自己哪裡錯。\u003C\u002Fli>\u003Cli>商業可用性成長很快，學界對 AGI 定義卻還沒統一。\u003C\u002Fli>\u003C\u002Ful>\u003Cp>這也是黃仁勳說法有意思的地方。他逼大家把問題講白。你是在談研究理想，還是在談經濟門檻？如果是後者，那討論就沒那麼玄。\u003C\u002Fp>\u003Cp>另外，這也解釋了 NVIDIA 為什麼樂見這種討論。Agent 越多，推理次數越多，工具呼叫越多，GPU 需求就越高。黃仁勳不是只在評論 AI。他也在描述一種更吃算力的軟體型態。\u003C\u002Fp>\u003Ch2>這場爭論的背景\u003C\u002Fh2>\u003Cp>AI 這幾年最大變化，不是模型會講話了。是模型開始接工具了。從訓練資料、推理成本，到 API 佈署方式，整個產業都在往「可執行」靠近。這讓 AGI 的討論變得更商業化，也更混亂。\u003C\u002Fp>\u003Cp>台灣開發者其實很容易感受到這件事。以前做 chatbot，重點是 prompt。現在做產品，重點是 workflow。你要管權限、資料來源、上下文長度、錯誤重試，還要顧伺服器成本。這些都不是哲學題，是工程題。\u003C\u002Fp>\u003Cp>所以我覺得，AGI 這個詞接下來會分裂成兩種用法。一種留在學術界。另一種留在產品圈。前者談通用能力。後者談能不能自己產值。兩邊都合理，但別混在一起講。\u003C\u002Fp>\u003Ch2>接下來該看什麼\u003C\u002Fh2>\u003Cp>接下來，最值得看的是 agent 能不能少靠人工。只要一個系統能穩定完成真實工作，像是客服分流、內部知識查詢、程式修補、報表整理，黃仁勳那種實用版 AGI 就會更有說服力。\u003C\u002Fp>\u003Cp>我的判斷很直接。明年大家討論的重點，可能不是「AGI 到了沒」。而是「你的團隊還要幾個人，才能把這套 AI 管住」。這個問題，比抽象定義更實際。\u003C\u002Fp>\u003Cp>如果你是開發者，現在最該看的不是神話。是工具鏈。誰能把模型、API、記憶、權限、監控串好，誰就比較接近下一波產品機會。說真的，這才是重點。\u003C\u002Fp>","黃仁勳說 AGI 可能已經到了，但他講的是實用版定義。真正的爭點，是 AI 到底要多像人，才配叫 AGI。","www.magzter.com","https:\u002F\u002Fwww.magzter.com\u002Fstories\u002Ftechnology\u002FTechlife-News\u002FNVIDIA-CHIEF-JENSEN-HUANG-SAYS-WEVE-ACHIEVED-AGI-BUT-WHAT-ON-EARTH-IS-IT",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1775132875223-ozgf.png","industry","zh","32c37b25-877e-465c-a20a-b1c9e26c7738",[17,18,19,20,21,22,23,24,25],"AGI","Jensen Huang","NVIDIA","AI agent","OpenAI","Anthropic","Google DeepMind","LLM","人工智慧",9,"2026-04-02T12:27:32.615208+00:00","2026-04-02T12:27:32.493+00:00",{"tags":30,"relatedLang":47,"relatedPosts":51},[31,33,34,37,39,41,43,45],{"name":21,"slug":32},"openai",{"name":25,"slug":25},{"name":35,"slug":36},"Nvidia","nvidia",{"name":17,"slug":38},"agi",{"name":22,"slug":40},"anthropic",{"name":24,"slug":42},"llm",{"name":20,"slug":44},"ai-agent",{"name":23,"slug":46},"google-deepmind",{"id":15,"slug":48,"title":49,"language":50},"jensen-huang-says-agi-arrived-did-it-en","Jensen Huang Says AGI Arrived. Did It?","en",[52,58,64,70,76,82],{"id":53,"slug":54,"title":55,"cover_image":56,"image_url":56,"created_at":57,"category":13},"600a41d7-99a2-48cf-b80e-b28061c65767","andes-technology-20b-risc-v-soc-shipments-zh","Andes RISC-V SoC 出貨破 200 億","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782734588433-8mkx.png","2026-06-29T12:02:32.954092+00:00",{"id":59,"slug":60,"title":61,"cover_image":62,"image_url":62,"created_at":63,"category":13},"383d45a7-2778-436c-902c-fb0d064bfe56","onchain-insurance-proof-institutional-tokenization-test-zh","鏈上保險證明才是機構代幣化的真正考題","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782729171879-ih4l.png","2026-06-29T10:32:25.181256+00:00",{"id":65,"slug":66,"title":67,"cover_image":68,"image_url":68,"created_at":69,"category":13},"e4d40a87-9823-4a96-a9a1-0da241daee68","dtcc-tokenization-link-stellar-zh","DTCC 接上 Stellar，XLM 站上新舞台","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782728271600-sddm.png","2026-06-29T10:17:27.929404+00:00",{"id":71,"slug":72,"title":73,"cover_image":74,"image_url":74,"created_at":75,"category":13},"10f14e61-67c3-4c5e-b561-371efdffb18f","framework-tokenization-ai-financing-fund-zh","Framework 把代幣化變融資","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782727404282-u4vv.png","2026-06-29T10:02:58.99285+00:00",{"id":77,"slug":78,"title":79,"cover_image":80,"image_url":80,"created_at":81,"category":13},"b19bc35b-9d90-4c63-94ab-c46bd759da81","microsoft-investor-relations-page-map-zh","Microsoft 投資人關係頁面地圖","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782726471089-sl2s.png","2026-06-29T09:47:23.941243+00:00",{"id":83,"slug":84,"title":85,"cover_image":86,"image_url":86,"created_at":87,"category":13},"e6695966-a6f4-4b71-ab89-cd61bc205d43","microsoft-190b-ai-capex-plan-msft-452-zh","Microsoft 1900億美元 AI 支出壓力測試 MSFT 452","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1782724684761-zzp0.png","2026-06-29T09:17:26.886615+00:00",[89,94,99,104,109,114,119,124,129,134],{"id":90,"slug":91,"title":92,"created_at":93},"ee073da7-28b3-4752-a319-5a501459fb87","ai-in-2026-what-actually-matters-now-zh","2026 AI 真正重要的事","2026-03-26T07:09:12.008134+00:00",{"id":95,"slug":96,"title":97,"created_at":98},"83bd1795-8548-44c9-9a7e-de50a0923f71","trump-ai-framework-power-speech-state-preemption-zh","川普 AI 框架瞄準電力、言論與州權","2026-03-26T07:12:18.695466+00:00",{"id":100,"slug":101,"title":102,"created_at":103},"ea6be18b-c903-4e54-97b7-5f7447a612e0","nvidia-gtc-2026-big-ai-announcements-zh","NVIDIA GTC 2026 重點拆解","2026-03-26T07:14:26.62638+00:00",{"id":105,"slug":106,"title":107,"created_at":108},"4bcec76f-4c36-4daa-909f-54cd702f7c93","claude-users-spreading-out-and-getting-better-zh","Claude 用戶更分散，也更會用","2026-03-26T07:22:52.325888+00:00",{"id":110,"slug":111,"title":112,"created_at":113},"bd903b15-2473-4178-9789-b7557816e535","openclaw-raises-hard-question-for-ai-models-zh","OpenClaw 逼問 AI 模型價值","2026-03-26T07:24:54.707486+00:00",{"id":115,"slug":116,"title":117,"created_at":118},"eeac6b9e-ad9d-4831-8eec-8bba3f9bca6a","gap-google-gemini-checkout-fashion-search-zh","Gap 把結帳搬進 Gemini","2026-03-26T07:28:23.937768+00:00",{"id":120,"slug":121,"title":122,"created_at":123},"0740e53f-605d-4d57-8601-c10beb126f3c","google-pushes-gemini-transition-to-march-2026-zh","Google 把 Gemini 轉換延到 2026 年 3…","2026-03-26T07:30:12.825269+00:00",{"id":125,"slug":126,"title":127,"created_at":128},"e660d801-2421-4529-8fa9-86b82b066990","metas-llama-4-benchmark-scandal-gets-worse-zh","Meta Llama 4 分數風波又擴大","2026-03-26T07:34:21.156421+00:00",{"id":130,"slug":131,"title":132,"created_at":133},"183f9e7c-e143-40bb-a6d5-67ba84a3a8bc","accenture-mistral-ai-sovereign-enterprise-deal-zh","Accenture 攜手 Mistral AI 賣主權 AI","2026-03-26T07:38:14.818906+00:00",{"id":135,"slug":136,"title":137,"created_at":138},"191d9b1b-768a-478c-978c-dd7431a38149","mistral-ai-faces-its-hardest-year-yet-zh","Mistral AI 迎來最硬的一年","2026-03-26T07:40:23.716374+00:00"]