[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-claude-corps-ai-training-into-jobs-zh":3,"article-related-claude-corps-ai-training-into-jobs-zh":30,"series-industry-ef5cc7c5-ea69-42aa-a487-714339bb08d8":73},{"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":22,"views":26,"created_at":27,"published_at":28,"topic_cluster_id":29},"ef5cc7c5-ea69-42aa-a487-714339bb08d8","claude-corps-ai-training-into-jobs-zh","Claude Corps 把 AI 訓練變工作","\u003Cp data-speakable=\"summary\">我拆 \u003Ca href=\"\u002Ftag\u002Fanthropic\">Anthropic\u003C\u002Fa> 的 \u003Ca href=\"\u002Ftag\u002Fclaude\">Claude\u003C\u002Fa> Corps，順手給你一份可直接複製的 fellowship 模板。\u003C\u002Fp>\u003Cp>我盯 AI 訓練計畫很久了，很多都怪得很一致：課上得很滿、證書發得很快、簡報做得很漂亮，最後回到現場還是那堆爛流程。非營利最缺的不是「知道 AI 很重要」，而是有人真的坐進去，把那個亂七八糟的 intake、表單、回信、整理、追蹤，一格一格拆掉。這個落差一直讓我很煩。\u003C\u002Fp>\u003Cp>所以我看到 Forbes 這篇 \u003Ca href=\"https:\u002F\u002Fwww.forbes.com\u002Fsites\u002Fmichaeltnietzel\u002F2026\u002F06\u002F18\u002Fanthropic-invests-150-million-to-launch-1000-claude-corps-fellowships\u002F\">Claude Corps 報導\u003C\u002Fa>時，第一個反應不是「哇好大手筆」，而是「終於有人把訓練跟工作綁在一起了」。Anthropic 不是只叫大家去學 AI，它是想把人塞進真實組織裡，讓 AI 變成可交付的工作，而不是課後感想。這件事我比較有感。\u003C\u002Fp>\u003Cp>報導裡寫得很直接：Anthropic 要砸 \u003Ca href=\"https:\u002F\u002Fwww.anthropic.com\u002F\">150 million 美元\u003C\u002Fa>，做 1,000 個 fellows，年薪 \u003Ca href=\"https:\u002F\u002Fwww.codepath.org\u002F\">85,000 美元\u003C\u002Fa>，放進最多 400 家美國非營利組織。這不是辦一場 webinar，這是把 staffing 當成 adoption 的一部分在做。\u003C\u002Fp>\u003Ch2>這不是課程，這是人力配置\u003C\u002Fh2>\u003Cblockquote>“Anthropic is investing $150 million to launch Claude Corps, a national fellowship program that will place young people in full-time jobs at various nonprofits around the country that want to use artificial intelligence more effectively in their work.”\u003C\u002Fblockquote>\u003Cp>翻譯一下就是：Anthropic 沒把 AI adoption 當知識問題，而是當人力問題。非營利組織不是缺一份 AI 指南，它們缺的是一個真的坐在現場的人，能把「用 AI」翻成「這個收件流程怎麼改、這封回信怎麼半自動、這個報告怎麼先草擬、哪些事絕對不能亂自動化」。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781799531675-ar3q.png\" alt=\"Claude Corps 把 AI 訓練變工作\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>我看過太多公司買工具，然後裝作很驚訝地說「怎麼沒人用」。原因通常不是工具爛，是缺一個懂現場的人，把工具接進日常工作。Claude Corps 的思路很簡單：好，那我們連人一起養。\u003C\u002Fp>\u003Cp>這裡最重要的是結構，不是標語。全職、進場、為期一年，這三個條件缺一個，整個東西就會歪掉。遠端會變成客服支援線，兼職會變成 side project，只有訓練沒有落地，就會變成履歷上一行字，對組織沒什麼用。\u003C\u002Fp>\u003Cp>實操寫法：如果你在做公司、基金會、學校或公部門的 AI 計畫，先別問「怎麼教大家用 AI」，先問「哪個工作值得安排一個人進去把 AI 用起來」。先定工作，再定課程。\u003C\u002Fp>\u003Cul>\u003Cli>人要靠近 workflow，不要關在訓練教室。\u003C\u002Fli>\u003Cli>每個人只負責一個 owner、一个 measurable problem。\u003C\u002Fli>\u003Cli>輸出要是 operational 的，不是只會說自己學到很多。\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>真正狠的是它把門檻打開了\u003C\u002Fh2>\u003Cp>Anthropic 和合作方說，申請者要年滿 18 歲、全職工作經驗少於兩年，\u003Ca href=\"\u002Fnews\u002Fkimi-k27-code-highspeed-mode-skips-benchmarks-zh\">還不\u003C\u002Fa>要求學位。這個設計比品牌名字更重要，因為它直接砍掉了很多假裝公平、實際上很會挑背景的門檻。那種「我們歡迎新人才」但最後只收名校、履歷漂亮、講話很像已經在這圈子裡混過的人，我看多了，真的很膩。\u003C\u002Fp>\u003Cp>CodePath 的 CEO Michael Ellison 在報導裡說，\u003Ca href=\"https:\u002F\u002Fwww.codepath.org\u002F\">他們刻意要做得非常 accessible\u003C\u002Fa>，而且希望第一批 fellows 反映更廣泛的人群。這不是空話，這是選才邏輯。\u003C\u002Fp>\u003Cp>我以前碰過一堆計畫，嘴上說要找新血，實際上還是在篩學歷、篩名聲、篩那種你一看就知道「這人以前一定常進會議室」的氣質。結果就是舊管道繼續複製舊管道。Claude Corps 如果真的照這個標準做，至少會打開另一條路：年輕、沒那麼多資歷包袱，但願意學、願意試、願意把一個怪流程磨到能用的人。\u003C\u002Fp>\u003Cp>而且這不只是理想主義。非營利裡的 AI 工作，很多根本不需要博士。它需要的是會聽、會寫、會測、會改，而且不會把第一版當神主牌的人。這種能力其實很可訓練。比起訓練人適應組織，訓練組織真的接受 AI，反而更難。\u003C\u002Fp>\u003Cp>實操寫法：如果你要做自己的 fellowship 或 apprenticeship，先把那些裝門面的條件拿掉。問「有沒有解題能力」「會不會寫清楚」「遇到新工具會不會自己摸出來」，不要先拿學位當篩子，除非那個職位真的非學位不可。\u003C\u002Fp>\u003Cul>\u003Cli>申請題目用白話，不要故作高深。\u003C\u002Fli>\u003Cli>看學習速度和溝通，不要只看 pedigree。\u003C\u002Fli>\u003Cli>薪資要夠真，不要做成慈善式實習。\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>這 150 million 不是只在付薪水\u003C\u002Fh2>\u003Cp>Anthropic 的 150 million 美元，當然不只是薪資。它還包含 program design、合作夥伴管理、訓練內容、媒合、支援系統，還有那些讓 1,000 個人不會一頭撞進 400 家非營利然後把事情搞亂的基礎工程。這就是很多計畫最愛偷懶的地方：大家都記得發布會，沒人預算水管。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781799513291-hjws.png\" alt=\"Claude Corps 把 AI 訓練變工作\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>報導裡提到的結構是 Anthropic、CodePath、Social Finance 三方合作。這分工其實很合理。Anthropic 負責\u003Ca href=\"\u002Fnews\u002Fmoonshot-ai-kimi-models-growth-zh\">模型\u003C\u002Fa>和資金，CodePath 負責人才訓練，Social Finance 負責把 fellowship 做得像個會運作的制度，而不是一個漂亮但脆的專案名稱。你要這東西活下來，就不能只靠一家公司自己硬撐。\u003C\u002Fp>\u003Cp>我自己幫團隊做內部 AI pilot 時，最常卡的也不是 demo，而是交接。誰負責 onboarding？誰確認非營利端的人真的能接手？誰決定 prototype 夠不夠成熟可以\u003Ca href=\"\u002Fnews\u002Fkimi-k27-whats-new-and-how-to-run-it-zh\">上線\u003C\u002Fa>？如果這些沒先講好，最後就是一堆半成品自動化，外加一個快被搞瘋的營運同仁。\u003C\u002Fp>\u003Cp>還有一點很現實：fellow 年薪 85,000 美元。這不是小事。薪資會決定這個角色到底像工作，還是像履歷裝飾。給太少，留下來的人少；給得像樣，才會吸引認真的人，也比較不會一堆表演型參與者。\u003C\u002Fp>\u003Cp>實操寫法：預算不要只算 stipend，要把整個生命週期算進去。訓練、督導、合作夥伴 onboarding、結案 offboarding，每一段都要有錢。你少算這些，最後就會得到一個很貴、但很空的實驗。\u003C\u002Fp>\u003Ch2>非營利才是 AI 的壓力測試\u003C\u002Fh2>\u003Cp>Anthropic 說 fellows 會幫非營利組織改善營運、推進 mission。這句話聽起來很大，但其實很對。非營利的條件通常最硬：人少、工具亂、行政支援薄、事情永遠比資源多。AI 如果連這裡都幫不上忙，那很多 hype 真的只是噪音。\u003C\u002Fp>\u003Cp>我反而覺得這不是從最容易的客戶開始。企業有預算、有 IT、有採購；非營利沒有這些，還得每天處理更雜的事。這種環境更能看出 AI 到底能不能真的省下重工，而不是只會製造新的管理負擔。\u003C\u002Fp>\u003Cp>但這裡也有陷阱。所謂「更好地用 AI」如果沒定義 use case，很容易變成空話。比較有用的版本會是：減少 donor 回信時間、整理 case notes、草擬 grant language、分類 incoming request、翻譯素材、協助志工排班。危險的版本就是某個沒待過現場的人喊一句「把全部都自動化」。\u003C\u002Fp>\u003Cp>實操寫法：先挑一個痛、重複、而且能量化的流程。不要從 mission statement 開始，先從 bottleneck 開始。\u003C\u002Fp>\u003Cul>\u003Cli>先找一個前後時間成本清楚的流程。\u003C\u002Fli>\u003Cli>敏感輸出一定保留人工審核。\u003C\u002Fli>\u003Cli>把改了什麼寫下來，讓組織在 fellowship 結束後還能接著跑。\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>分批上線，代表他們知道第一版一定會歪\u003C\u002Fh2>\u003Cp>Forbes 寫，第一批申請在 7 月 17 日截止，100 位 fellows 會在 2026 年 10 月開始；之後還會在 2027 年 1 月和 8 月再開兩批。這種分批節奏我其實很喜歡，因為它表示他們沒有假裝第一版就會完美。\u003C\u002Fp>\u003Cp>Daniela Amodei 在 \u003Ca href=\"https:\u002F\u002Fapnews.com\u002F\">Associated Press\u003C\u002Fa> 裡說過一句我很認同的話：\u003Cblockquote>“We’re hoping it’s a good idea that can take root and that other people can build on and learn from.”\u003C\u002Fblockquote>這句聽起來很務實。第一批不是產品本體，第一批是測試。\u003C\u002Fp>\u003Cp>我真的很受不了那種把 pilot 說得像紀念碑的 AI 計畫。發表時聲勢很大，六個月後才發現現場根本吃不下，然後默默改 scope。分批上線至少還有一個好處：你可以根據第一波的問題，去改 curriculum、媒合方式、支援模型和 host expectations。\u003C\u002Fp>\u003Cp>實操寫法：如果你也要辦類似計畫，務必在 cohort 之間留空檔。第一批負責找 bug，第二批才拿來修正後擴大。不要把 bug 一起 scale。\u003C\u002Fp>\u003Ch2>這件事真正講的是 AI 怎麼被分發\u003C\u002Fh2>\u003Cp>Claude Corps 最有意思的地方，是它把 AI access 當成可以透過人來分發，不只是透過產品更新。這點很多公司都漏看了。他們以為 adoption 會在模型更強那刻自動發生。其實通常不是。通常是某個離現場很近的人，知道怎麼把工具塞進真實工作裡， adoption 才開始動。\u003C\u002Fp>\u003Cp>所以這個計畫的價值，可能不只在非營利。如果它跑得順，它會變成一種模板：AI 廠商、基金會、大學、公民組織，都可以把 skilled operator 放進資源不足的機構裡。不是那種做完簡報就走的顧問，而是會留下實際能力的人。\u003C\u002Fp>\u003Cp>我不會裝作這能解決所有事。它一定會有媒合失敗、use case 誇大、還有一些 fellows 進去後才發現非營利營運根本是另一種混亂。但比起再做一個叫大家「擁抱 AI」然後拍拍屁股走人的活動，我寧願看到一個真的把人放進流程裡的方案。\u003C\u002Fp>\u003Cp>實操寫法：別只想「訓練使用者」，改成想「嵌入翻譯者」。這是我看完後一直卡在腦中的職位概念。那個人要同時懂工具和 workflow，還要有權限推動工作方式改變。\u003C\u002Fp>\u003Ch2>可抄的模板\u003C\u002Fh2>\u003Cpre>\u003Ccode># AI Fellowship Program Template for Nonprofits\n\n## Program name\n[Your program name]\n\n## Purpose\nFund and place early-career fellows inside nonprofit organizations to apply AI to real operational work.\n\n## Program goals\n- Reduce repetitive administrative work\n- Improve response time on core workflows\n- Build lasting AI capability inside host organizations\n- Train fellows to use AI responsibly in mission-driven settings\n\n## Fellowship structure\n- Duration: 12 months\n- Format: Full-time, in-person\n- Cohort size: [number]\n- Host organizations: [number]\n- Compensation: [$ amount] per year\n- Start dates: [date 1], [date 2], [date 3]\n\n## Eligibility\nApplicants must:\n- Be 18 or older\n- Have less than 2 years of full-time work experience\n- Be authorized to work in [country]\n- Be comfortable using AI tools\n- Not require a college degree unless the role truly needs one\n\n## Selection criteria\nPrioritize applicants who show:\n- Clear writing and communication\n- Fast learning and curiosity\n- Comfort with ambiguity\n- Respect for human review in sensitive work\n- Evidence of solving practical problems\n\n## Host organization criteria\nSelect nonprofits that:\n- Have a real workflow bottleneck\n- Can provide a day-to-day manager\n- Will commit staff time to onboarding and feedback\n- Want to keep the improvements after the fellowship ends\n\n## Fellow responsibilities\nFellows may:\n- Map workflows and identify automation opportunities\n- Draft and test prompts, templates, and SOPs\n- Build lightweight AI-assisted processes\n- Train staff on approved tool use\n- Document what works and what should not be automated\n\n## Guardrails\n- No fully autonomous decisions for sensitive cases\n- Human review required for external communications, legal, financial, and beneficiary-impacting outputs\n- Track data privacy and access rules\n- Maintain logs of AI-assisted changes\n\n## Support model\nEach fellow should have:\n- One program manager\n- One host-site supervisor\n- One technical mentor\n- One monthly review on progress and risks\n\n## Success metrics\nMeasure:\n- Hours saved per week\n- Workflow turnaround time\n- Staff adoption rate\n- Number of reusable templates\u002Fprocesses created\n- Post-fellowship continuation rate\n\n## Offboarding\nBefore the fellow exits:\n- Document workflows and owners\n- Train a permanent staff member\n- Package prompts, templates, and SOPs\n- Identify which tools to keep, replace, or retire\n\n## Simple application prompt\nDescribe one workflow you would improve in a nonprofit, what you would automate first, and how you would keep humans in control of the final decision.\n\n## Host application prompt\nDescribe the bottleneck you want solved, the staff member who will supervise the fellow, and what success looks like after 90 days.\u003C\u002Fcode>\u003C\u002Fpre>\u003Cp>這份模板是我根據 Forbes 報導重組出來的版本，不是 Anthropic 原文措辭。原始數字、合作方和計畫輪廓來自 \u003Ca href=\"https:\u002F\u002Fwww.forbes.com\u002Fsites\u002Fmichaeltnietzel\u002F2026\u002F06\u002F18\u002Fanthropic-invests-150-million-to-launch-1000-claude-corps-fellowships\u002F\">Michael T. Nietzel 在 Forbes 的文章\u003C\u002Fa>；我這邊補的是可直接拿去改成你自己版本的 structure。\u003C\u002Fp>\u003Cp>來源：\u003Ca href=\"https:\u002F\u002Fwww.forbes.com\u002Fsites\u002Fmichaeltnietzel\u002F2026\u002F06\u002F18\u002Fanthropic-invests-150-million-to-launch-1000-claude-corps-fellowships\u002F\">Forbes\u003C\u002Fa>、\u003Ca href=\"https:\u002F\u002Fwww.anthropic.com\u002F\">Anthropic\u003C\u002Fa>、\u003Ca href=\"https:\u002F\u002Fwww.codepath.org\u002F\">CodePath\u003C\u002Fa>、\u003Ca href=\"https:\u002F\u002Fapnews.com\u002F\">AP News\u003C\u002Fa>。文中的框架拆解、實操建議和模板重寫是我自己的整理，原始報導的事實部分我有對照來源後再下筆。","我拆 Anthropic Claude Corps 的打法，順手給你一份可直接複製的非營利 AI fellowship 模板。","www.forbes.com","https:\u002F\u002Fwww.forbes.com\u002Fsites\u002Fmichaeltnietzel\u002F2026\u002F06\u002F18\u002Fanthropic-invests-150-million-to-launch-1000-claude-corps-fellowships\u002F",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781799531675-ar3q.png","industry","zh","7592a176-05e4-44e9-b76f-b2c8a5364dcc",[17,18,19,20,21],"AI fellowship","nonprofit operations","workforce development","Claude Corps","apprenticeship",[23,24,25],"Claude Corps 把 AI adoption 當人力與流程問題，不是單純課程問題。","低門檻、全職、進場、可衡量的設計，比漂亮宣傳更重要。","最有價值的是可直接複製的 fellowship 模板，能改成你自己的計畫。",0,"2026-06-18T16:17:58.590726+00:00","2026-06-18T16:17:58.565+00:00","29fa8a72-a8a8-473e-975c-3991ae762f60",{"tags":31,"relatedLang":32,"relatedPosts":36},[],{"id":15,"slug":33,"title":34,"language":35},"anthropic-claude-corps-ai-nonprofit-fellowships-en","Anthropic’s Claude Corps turns AI training into jobs","en",[37,43,49,55,61,67],{"id":38,"slug":39,"title":40,"cover_image":41,"image_url":41,"created_at":42,"category":13},"f2177583-b1dd-4de8-8adf-6754d7c8f9c2","hermes-agent-learning-memory-gateways-zh","Hermes Agent 的 5 個關鍵能力","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781867872053-6r9l.png","2026-06-19T11:17:23.11904+00:00",{"id":44,"slug":45,"title":46,"cover_image":47,"image_url":47,"created_at":48,"category":13},"39cf67a3-ce5a-4a4c-9322-a42a6e6476c8","rust-built-different-update-server-admins-zh","Rust 6 月更新讓伺服器管理員頭痛","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781851683502-ql4f.png","2026-06-19T06:47:36.493495+00:00",{"id":50,"slug":51,"title":52,"cover_image":53,"image_url":53,"created_at":54,"category":13},"7088e08a-2624-4ee5-bc20-136f59f3a627","openai-ipo-prep-policy-hiring-play-zh","OpenAI IPO 前先把政策變成招募","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781842703362-p8sd.png","2026-06-19T04:17:56.622121+00:00",{"id":56,"slug":57,"title":58,"cover_image":59,"image_url":59,"created_at":60,"category":13},"8e153a8b-a625-4797-a8cb-b7fbc7273810","openai-right-to-hire-dean-ball-policy-power-zh","OpenAI 延攬 Dean Ball 是對的：政策權力已是 AI 核心戰場","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781841764947-t4hd.png","2026-06-19T04:02:18.251835+00:00",{"id":62,"slug":63,"title":64,"cover_image":65,"image_url":65,"created_at":66,"category":13},"7d221962-c85f-4776-af7f-cedd921d67fb","rust-rolling-release-model-right-tradeoff-zh","Rust 的滾動式支援，才是團隊真正該接受的取捨","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781834562661-8eed.png","2026-06-19T02:02:18.140322+00:00",{"id":68,"slug":69,"title":70,"cover_image":71,"image_url":71,"created_at":72,"category":13},"5d19eff9-8a4f-4ca4-ac8a-e793ebadf7f4","cloudflare-director-vote-pressure-ai-push-zh","Cloudflare 面臨董事投票壓力","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781830969389-5tea.png","2026-06-19T01:02:25.790674+00:00",[74,79,84,89,94,99,104,109,114,119],{"id":75,"slug":76,"title":77,"created_at":78},"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":80,"slug":81,"title":82,"created_at":83},"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":85,"slug":86,"title":87,"created_at":88},"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":90,"slug":91,"title":92,"created_at":93},"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":95,"slug":96,"title":97,"created_at":98},"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":100,"slug":101,"title":102,"created_at":103},"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":105,"slug":106,"title":107,"created_at":108},"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":110,"slug":111,"title":112,"created_at":113},"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":115,"slug":116,"title":117,"created_at":118},"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":120,"slug":121,"title":122,"created_at":123},"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"]