[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-dara-think-tanks-ai-trust-zh":3,"article-related-dara-think-tanks-ai-trust-zh":33,"series-industry-1787c1f6-5b34-4ddc-9eb9-4ec10e898711":77},{"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":25,"views":29,"created_at":30,"published_at":31,"topic_cluster_id":32},"1787c1f6-5b34-4ddc-9eb9-4ec10e898711","dara-think-tanks-ai-trust-zh","DARA把智庫 AI 透明化","\u003Cp data-speakable=\"summary\">iNNOV8 的 DARA 是一個有人類監督的 AI 研究助手，重點在測試智庫工作裡的揭露、作者歸屬與信任規則。\u003C\u002Fp>\u003Cp>這東西蠻有意思。\u003Ca href=\"https:\u002F\u002Fonthinktanks.org\u002Farticles\u002Fdara-trust-and-the-next-stage-of-ai-in-thinktanks\u002F\" target=\"_blank\" rel=\"noopener\">On Think Tanks\u003C\u002Fa> 先寫到這個案子，\u003Ca href=\"https:\u002F\u002Finnov8.iq\u002F\" target=\"_blank\" rel=\"noopener\">iNNOV8\u003C\u002Fa> 再把 DARA 推上檯面。它不是單純做一個聊天\u003Ca href=\"\u002Fnews\u002Fveritas-robot-policy-visual-verification-zh\">機器人\u003C\u002Fa>，而是直接拿智庫研究\u003Ca href=\"\u002Fnews\u002Fkimi-k27-review-copyable-coding-playbook-zh\">流程\u003C\u002Fa>當測試場。\u003C\u002Fp>\u003Cp>時間點也很巧。文章提到的主題是 \u003Cstrong>Between Knowledge and Algorithm: Generative AI in the Think Tank Environment\u003C\u002Fstrong>。而 2026 年的 \u003Ca href=\"https:\u002F\u002Fonthinktanks.org\u002Fott-conference\u002F\" target=\"_blank\" rel=\"noopener\">OTT Conference\u003C\u002Fa> 在拉巴特談的核心，就是信任。講白了，這不是工具展示，是制度測試。\u003C\u002Fp>\u003Ctable>\u003Cthead>\u003Ctr>\u003Cth>項目\u003C\u002Fth>\u003Cth>內容\u003C\u002Fth>\u003C\u002Ftr>\u003C\u002Fthead>\u003Ctbody>\u003Ctr>\u003Ctd>專案\u003C\u002Ftd>\u003Ctd>DARA, the Dynamic Analysis and Research Assistant\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>地點\u003C\u002Ftd>\u003Ctd>Sulaymaniyah, Iraq\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>首篇論文\u003C\u002Ftd>\u003Ctd>Between Knowledge and Algorithm: Generative AI in the Think Tank Environment\u003C\u002Ftd>\u003C\u002Ftr>\u003Ctr>\u003Ctd>會議脈絡\u003C\u002Ftd>\u003Ctd>OTT Conference 2026 in Rabat\u003C\u002Ftd>\u003C\u002Ftr>\u003C\u002Ftbody>\u003C\u002Ftable>\u003Ch2>DARA 先把 AI 用法攤開\u003C\u002Fh2>\u003Cp>多數智庫早就碰 AI 了。研究員拿它掃文獻、寫大綱、摘要、翻譯。公關團隊拿它改新聞稿。營運團隊拿它排流程。問題是，很多組織的使用速度，比內規快太多。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781684279103-4af9.png\" alt=\"DARA把智庫 AI 透明化\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>DARA 的做法剛好反過來。iNNOV8 說，這個助手會在內部監督下提出研究題目、規劃方法、起草文章。它也會清楚標示 AI 產出。可行的話，還會把方法和 prompts 一起揭露。\u003C\u002Fp>\u003Cp>這種透明度，比「有沒有用 AI」更重要。因為真正麻煩的，不是工具本身，而是誰知道它參與了多少。對智庫來說，信任不是口號，是流程設計。\u003C\u002Fp>\u003Cul>\u003Cli>AI 可以幫忙轉錄、搜尋、起草、翻譯。\u003C\u002Fli>\u003Cli>AI 不該取代訪談、田野、價值判斷。\u003C\u002Fli>\u003Cli>只要 AI 進入最後稿件，就該揭露。\u003C\u002Fli>\u003Cli>最終論點還是要由人類研究員負責。\u003C\u002Fli>\u003C\u002Ful>\u003Ch2>智庫賣的其實是信任\u003C\u002Fh2>\u003Cp>在 OTT Conference 上，AI 幾乎場場都會被提到。可是真正的焦點，不是某個模型多強，而是機構信用會不會被稀釋。當機器可以比人更快產出政策簡報、利害關係人地圖、研究草稿，智庫到底還剩什麼價值？\u003C\u002Fp>\u003Cp>Erica Schoder 在會議上的說法很直白。AI 很會做「complexity work」。它能快速處理大量資料。可是什麼重要、什麼可以接受、什麼後果要自己扛，還是人類決定。這條線如果畫不清楚，智庫就會把自己做成內容工廠。\u003C\u002Fp>\u003Cblockquote>“AI can do extraordinary ‘complexity work,’ such as processing large quantities of information quickly and helping small organisations extend their capacity.” — Erica Schoder\u003C\u002Fblockquote>\u003Cp>我覺得這句很準。若一家智庫只拼產量，AI 會直接把它打趴。若它賣的是判斷、責任、關係網和政治理解，AI 就只是工具。DARA 的價值在於，它把信任放進流程，而不是放在出版後的公關說明裡。\u003C\u002Fp>\u003Cp>這篇文章其實把三件事切開了。能不能用 AI，是一題。要不要揭露，是一題。最後能不能出版，是第三題。很多組織老是把這三題混在一起，然後內控就炸掉。\u003C\u002Fp>\u003Ch2>智庫其實早就有規則素材\u003C\u002Fh2>\u003Cp>這不是第一次有人在 On Think Tanks 上談這題。2023 年，他們就寫過在智庫裡用 \u003Ca href=\"https:\u002F\u002Fopenai.com\u002Fchatgpt\" target=\"_blank\" rel=\"noopener\">ChatGPT\u003C\u002Fa> 的實務建議，重點很簡單：AI 內容一定要人工覆核。那時候看起來像提醒，現在看起來像基本功。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781684276480-16ix.png\" alt=\"DARA把智庫 AI 透明化\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>到了 2024 年，Enrique Mendizabal 又開始想像 AI 助手會嵌進研究、募資、策略和政策溝通。Aidan Muller 談的是準備度。Joscha Wirtz 強調的是 intentionality，不要只是追 FOMO。Tony Bader 甚至提過內部的「AI constitution」。\u003C\u002Fp>\u003Cp>這些想法現在慢慢收斂成一份很實際的清單。智庫要先知道員工怎麼偷偷用 AI，再寫出大家真的看得懂的規則。接著要分清楚輔助和取代。最後還要把品質檢查攤給讀者看。\u003C\u002Fp>\u003Cul>\u003Cli>先盤點組織內部已經存在的 AI 用法。\u003C\u002Fli>\u003Cli>把允許、禁止、揭露寫成短規則。\u003C\u002Fli>\u003Cli>明確區分輔助與取代。\u003C\u002Fli>\u003Cli>用方法註記、metadata、資金來源、編輯審核來補信任。\u003C\u002Fli>\u003C\u002Ful>\u003Cp>還有一個常被忽略的點。負責任的 AI 使用不是免費的。它需要訓練、內網、權限控管、編輯人力。這些都不是附加品，而是讓政策研究不失真的基礎建設。\u003C\u002Fp>\u003Ch2>地點比軟體更重要\u003C\u002Fh2>\u003Cp>DARA 來自 Sulaymaniyah，這點很關鍵。它不是華盛頓、倫敦或布魯塞爾的明星實驗室。它來自一個區域性場域。這代表它面對的語言、資源和曝光條件都不一樣。\u003C\u002Fp>\u003Cp>小型或區域型智庫的風險也更明顯。它們的資料可能比較少被大型模型看見。它們的語言覆蓋率也可能較低。它們的信用訊號，對自動化系統來說也沒那麼容易辨識。\u003C\u002Fp>\u003Cp>但 AI 也能幫忙。它可以翻譯、整理、擴充產能，讓小團隊接觸到原本碰不到的受眾。前提是，它不能偷偷把在地脈絡磨平。只要有清楚監督，AI 就是在幫忙，不是在搶走研究員的位置。\u003C\u002Fp>\u003Cp>我會把 DARA 的訊號解讀成一句話：智庫下一\u003Ca href=\"\u002Fnews\u002Fsolana-unchained-stage-2-presale-007-enterprise-sdk-zh\">階段\u003C\u002Fa>要比的，不是誰先上 AI，而是誰先把責任寫清楚。誰負責、誰揭露、誰審稿，這三件事會決定讀者還信不信你。\u003C\u002Fp>\u003Ch2>智庫 AI 的下一步很現實\u003C\u002Fh2>\u003Cp>如果你在做政策研究，現在最該問的不是「要不要用 AI」。那題太慢了。真正該問的是，哪些環節可以用，哪些一定要人做，哪些地方必須公開說明。\u003C\u002Fp>\u003Cp>DARA 提供的是一個範本。它不把 AI 包裝成神隊友，也不把 AI 當成洪水猛獸。它只是在說，研究流程可以被設計得更透明。這種做法很土，但很有效。\u003C\u002Fp>\u003Cp>接下來我會觀察兩件事。第一，更多智庫會不會跟進揭露 AI 參與程度。第二，讀者會不會開始要求方法、prompt 和人工審核紀錄。這兩件事如果真的發生，智庫圈的 AI 規則就不會再只是內部備忘錄了。\u003C\u002Fp>","iNNOV8 的 DARA 用人類監督的 AI 研究流程，測試智庫如何揭露作者、方法與 AI 參與程度，讓信任成為政策研究的核心規則。","onthinktanks.org","https:\u002F\u002Fonthinktanks.org\u002Farticles\u002Fdara-trust-and-the-next-stage-of-ai-in-think-tanks\u002F",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781684279103-4af9.png","industry","zh","f28e0b4a-13db-4c25-85e5-6ac2dd37adba",[17,18,19,20,21,22,23,24],"DARA","智庫","人工智慧","政策研究","AI 透明度","信任","On Think Tanks","iNNOV8",[26,27,28],"DARA 把 AI 研究流程公開化，重點是揭露而不是遮掩。","智庫的核心資產是信任，AI 只能當工具，不能取代責任。","區域型智庫也能用 AI 擴充產能，但前提是有清楚監督與方法揭露。",0,"2026-06-17T08:17:31.003261+00:00","2026-06-17T08:17:30.987+00:00","caa87b65-9bbc-46fe-bba8-4f4158dd2d8b",{"tags":34,"relatedLang":36,"relatedPosts":40},[35],{"name":19,"slug":19},{"id":15,"slug":37,"title":38,"language":39},"dara-think-tanks-ai-trust-en","DARA shows how think tanks can use AI with trust","en",[41,47,53,59,65,71],{"id":42,"slug":43,"title":44,"cover_image":45,"image_url":45,"created_at":46,"category":13},"2a151488-09f9-4aa8-a654-3f1d9d7e159c","china-ai-open-source-efficiency-global-sales-zh","中國 AI 轉向：開源、效率、出海","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781702271555-i3q3.png","2026-06-17T13:17:25.59471+00:00",{"id":48,"slug":49,"title":50,"cover_image":51,"image_url":51,"created_at":52,"category":13},"4a2fbd38-b5c2-4590-9d4b-87f39f95ab9c","ergo-hestia-pricing-time-to-market-databricks-zh","ERGO Hestia 4 招縮短定價上線","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781697768906-9krk.png","2026-06-17T12:02:22.440161+00:00",{"id":54,"slug":55,"title":56,"cover_image":57,"image_url":57,"created_at":58,"category":13},"0cf56d85-887b-4fb1-8589-046da6513d26","openai-oracle-universal-credits-enterprise-buying-zh","OpenAI 進 Oracle 企業採購圈","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781696892976-sx90.png","2026-06-17T11:47:35.092555+00:00",{"id":60,"slug":61,"title":62,"cover_image":63,"image_url":63,"created_at":64,"category":13},"dd3d240a-0f53-49a4-90a5-cac17171f3fd","managed-chatgpt-access-policy-layers-zh","4 層規範決定企業版 ChatGPT 可怎麼用","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781695973066-pbtw.png","2026-06-17T11:32:17.633521+00:00",{"id":66,"slug":67,"title":68,"cover_image":69,"image_url":69,"created_at":70,"category":13},"c826a181-b373-4a9e-a494-1f8f4bc86c3c","openai-service-terms-app-risk-users-zh","OpenAI 服務條款把第三方 App 風險留給使用者","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781695063951-v71m.png","2026-06-17T11:17:21.223004+00:00",{"id":72,"slug":73,"title":74,"cover_image":75,"image_url":75,"created_at":76,"category":13},"7b6bec1f-4f42-4b60-a72d-027bf95a36e7","anthropic-fable-shutdown-own-your-models-zh","Fable 停用逼你把模型收回來","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1781687002361-q7fl.png","2026-06-17T09:02:52.16704+00:00",[78,83,88,93,98,103,108,113,118,123],{"id":79,"slug":80,"title":81,"created_at":82},"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":84,"slug":85,"title":86,"created_at":87},"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":89,"slug":90,"title":91,"created_at":92},"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":94,"slug":95,"title":96,"created_at":97},"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":99,"slug":100,"title":101,"created_at":102},"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":104,"slug":105,"title":106,"created_at":107},"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":109,"slug":110,"title":111,"created_at":112},"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":114,"slug":115,"title":116,"created_at":117},"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":119,"slug":120,"title":121,"created_at":122},"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":124,"slug":125,"title":126,"created_at":127},"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"]