AI行業如何“反內卷”？

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"從技術的發明到落地的“最後一公里”問題，是最具社會價值的命題。在人口紅利逐漸消退的今天，各行各業都面臨着人力成本居高不下，流程效率亟待待提高，拓客面不斷變窄的影響，AI的興起帶來了一股轉型的新風，卻仍與人們最初的期待存在着差距。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"11月23日，"},{"type":"link","attrs":{"href":"https:\/\/conference.idea.edu.cn\/%E5%A4%A7%E4%BC%9A%E6%97%A5%E7%A8%8B","title":"xxx","type":null},"content":[{"type":"text","text":"粵港澳大灣區數字經濟研究院"}]},{"type":"text","text":"（International Digital Economy Academy，簡稱“IDEA”）CTO Labs 總負責人幺寶剛與氪信科技創始人朱明傑、 智加科技中國區總經理容力、 翼方健數首席科學家張霖濤等AI產業界代表者共同開啓AI產業落地論壇，分享了各自在行業深耕的經歷和思考。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"以下內容由InfoQ整理自現場速記，經過不改變原意的刪減。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"主持人："},{"type":"text","text":"人工智能產業化的時代已經到來。但如何做“可落地”的AI，真正體現出AI的社會效益和商業價值？在這個論壇上我們會邀請幾位來自不同領域的行業大咖，分享他們在AI落地過程的經驗和心得，共同探討AI產業化的挑戰、道路和前景。有請："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"    IDEA CTO Labs 總負責人幺寶剛"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"    翼方健數首席科學家張霖濤"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"    智加科技中國區總經理容力"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"    氪信科技創始人朱明傑"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":"大家早上好！我在IDEA是負責CTO Labs建設的，所以我對AI產業化一直非常關注。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"昨天Harry也提到過，我們既要仰望星空又要腳踏實地，所以選擇了一個很腳踏實地的話題。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"首先我介紹一下在座的三位嘉賓："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"朱明傑博士是我們"},{"type":"text","marks":[{"type":"strong"}],"text":"CTO 工作室第一批入駐的成員"},{"type":"text","text":"，氪信科技的創始人，畢業於科大少年班是MIT Take Review評選出的35歲以下創新的35人之一、上海市青年拔尖人才，氪信也是首批與IDEA成立CTO Labs的企業，我們也很期待接下來把這個工作室一起建設好。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"張霖濤博士，畢業於北京大學和普林斯頓大學，也是IEEE Fellow，在翼方健數擔任首席科學家。翼方健數是我們昨天剛剛宣佈的第二批工作室成員之一，也非常感謝霖濤博士的支持。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"容力博士，畢業於清華大學和滑鐵盧大學，在人工智能領域有20多年的實踐經驗和經歷。他現在是智加科技的工程高級副總監和中國區總經理，也是自動駕駛領域，自動駕駛領域一直是我們很關注的賽道，也非常希望未來有這個領域的CTO工作室成立，也期待跟容力博士一起探討未來的合作機會。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"接下來請幾位嘉賓先從他們自身的業務出發，來介紹一下他們怎麼定義可落地的AI。從明傑博士開始。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"朱明傑："},{"type":"text","text":"AI要落地，首先行業的痛點要足夠明顯。氪信科技是一家AI公司，面向落地的行業主要是金融行業。現在中國尤其是銀行，資產規模前15位的銀行應該大部分都是我們的客戶。我們是幫助他們做深度數字化轉型的合作伙伴。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"關於AI可落地這件事，首先要回答：這個行業或者這個場景爲什麼需要AI？我們就拿銀行來舉例子。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"2010年以前，銀行業是跑馬圈地階段，是一個人口紅利時代，遍地是客戶。一個銀行要發展很簡單，開網點，多去超市門口發金龍魚。那時候肯定不需要什麼AI，只要多僱人，多開網點就可以。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"2010年以後，是來自於互聯網的創新的良性啓發階段，大家都去開網點，競爭有點飽和了。還有像螞蟻、微衆包括很多互聯網金融創新的挑戰。金融行業裏大家爲了爭奪開戶，必須要學習互聯網的服務方式，要圍着客戶去轉。反映到銀行身上，就是學習像餘額寶一樣做一些創新。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"現在行業信息基礎建設進程普遍起來了，這幾年銀行可能真的很痛苦，因爲用戶今天不缺金融服務，想做個轉帳和支付交易，都可以不需要銀行。而且現在風險越來越高，監管越來越嚴，這就是內卷。內卷以後要逼着大家找一個能放大它的生產力的大殺器，那就是AI。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" 今天你再問一家銀行需不需要AI，他一定說需要的，因爲痛點真的很強烈的。再舉個具體的例子：去年中國破獲的電信網絡詐騙案件32萬起，今年上半年同比增長了60%，這個工作量和精準度不是人力能包攬的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在這種場景上，我們也可以看到"},{"type":"text","marks":[{"type":"strong"}],"text":"金融行業很像中國社會的代表縮影——從人口紅利開始往工程師紅利。要藉助工程師的力量、AI的力量做這件事情。"},{"type":"text","text":"我們做的也就是落地的事情。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":" 有一個跟進的問題：我在物流領域工作過一段時間，一個大的問題是我們去哪裏找機會點。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"當時我們有一個非常簡單的原則，就是哪兒人多就去哪兒找機會點。物流行業就是人力成本，把人力降下來。這是立竿見影能看到效果的。 那金融領域怎麼找機會點呢？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"朱明傑："},{"type":"text","text":"我們從2015年底開始創業面向金融行業時就提到過：真愛只能用錢表達。我們做的事情一定是客戶願意買單的，願意買單就是真實的痛點。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"就像我剛纔講的，現在電信網絡詐騙，它真的非常痛。因爲這件事情如果沒有做好，用戶會來投訴，監管又要罰款，很多時候有這種客觀的訴求存在，但又不可能靠業務人員一筆一筆查到是怎麼回事，倒逼着他要想這件事情。這裏面來自業務的需求在，我們就去做這件事情。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"我們也定義了一個概念：有效AI應用場景。"},{"type":"text","text":"要甄別出無效或者是僞應用場景，以免干擾我們的資源和經歷。很幸運的，氪信過去這五年裏，跟中國頭部的金融機構不斷探索、探討哪些領域是剛需的。我們梳理了大概24個有效AI應用場景，包括信用反欺詐、反洗錢、可疑交易識別、智能營銷。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"這裏面有很多細分領域，感興趣的同志可以關注我們公司後續一些產品介紹，謝謝大家！"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":"謝謝明傑博士。容力博士，從你們業務角度介紹一下什麼叫可落地的AI。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"   "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"容力："},{"type":"text","text":"我們公司叫智加科技，是做重卡幹線物流上的自動駕駛。所謂幹線物流是指長途運輸，主要跑高速公路，做自動駕駛，重卡是指我國拉着集裝箱的大卡車。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"談到落地應用，我們做了這幾年，深有體會。因爲我們做的不是一個研究的項目，是要應用在重卡上的項目。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"基本總結，一個可以落地的人工智能技術，在我們看來有三個點很重要：一是選中的應用一定要解決一個什麼痛點或者是一個什麼問題。可以是一個經濟問題，也可以是一個社會問題，也可以既是經濟問題也是社會問題。重卡的運輸，是經濟+社會的問題。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"從經濟上來講，降本增效。未來如果能夠實現無人駕駛，卡車司機這種繁重的工作，而且工資也比較高，它可以達到很高的降本增效效應。社會效應也是，如同以前織布機替代紡織工人一樣，帶來的是產業上的變化，社會效應也很大。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"二是要可以應用。因爲社會上痛點很多，問題很多，但不是所有的問題能在看得見的未來通過技術解決的。所以我們要選擇一個技術能夠落地解決的問題。比如現在腦機接口，不是說不能解決，十年看不到最後的應用成果。包括要探索人類的下一次移民的地方。這個問題很嚴重，看過《三體》的都知道，很多問題。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"但我們現在作爲一個公司，看中的是在看得見的三五年時間裏能解決。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"三是Scalable。選擇一個AI能夠解決社會或者經濟問題，而且可應用，還得Scalable，這樣才能掙更多錢。我們所看中的物流市場，是萬億級的市場，市場規模非常大，國民經濟將近15%GDP花在物流運輸上。如果對物流運輸帶來降本增效的作用，哪怕是幾個百分點，都是巨大的經濟效應。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":"下面有請張霖濤博士爲我們分享。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"   "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"張霖濤："},{"type":"text","text":"首先非常感謝能有這個機會跟大家分享一下我們在AI、數據方面做的一些工作，非常榮幸能夠加入CTO 工作室，來和IDEA合作，我們也Looking forward很多成果。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們公司叫翼方健數，做隱私計算賽道。公司已經有五年曆史。最近幾年隱私計算是大家比較關注的行業，我們也希望在這方面爲AI助力。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們公司跟其他一些隱私公司不太一樣的情況是，我們公司比較專注於醫療健康、生物信息等等行業。原因很簡單，因爲我們公司成立得比較早，當時其他行業都還在買數據。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"所謂可落地的AI，可能跟金融等行業不太一樣。比如金融，大家能夠看到這個數據的價值，比如能獲多少客，或者我的風險降低了多少個點。在有些數據使用得不是特別好的行業，比如醫療健康，這是非常傳統的行業，非常難啃的骨頭。怎麼樣把這個技術落地，是非常有挑戰的事情。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"醫療行業有非常大的潛力，我們稱之爲有很高勢能的行業。這個地方有很大的勢能，但中間有一個壩，怎麼能敲掉其中一些石頭，把這個水流下來一點，就會產生很大的經濟效益，"},{"type":"text","marks":[{"type":"strong"}],"text":"但你說上來就能把整快山移開，這是不太現實的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們在這個行業的思考是，用AI的能力幫助醫生或者幫助一些其他的醫療患者。這個功能必須要有醫院會買單或者對醫生有實實在在的價值。比如剛纔說的診療醫生，如果我能做一個醫生，非常了不起。你想做一個AI醫生，需要診療模型。這個診療模型的意思是根據你的症狀就知道得了什麼病，或者根據這些病，要得到什麼藥，要建一個大的AI系統。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"但這個事是非常難做的，包括IBM做沃森，Deep Mind做這些，都想一下子做出來。在實際應用中，你說“我想跟醫生說我有一個東西能代替你。”這是很難做的。我們想做一些醫生能認可的，像病例自控。比如你做了一個診療，這個人得了胃病，但你發現你的病例上診斷都是肺部的，肺的X光等等，要麼是病例寫錯了，要麼是當時的診斷有些問題。對於醫院病例科都是非常有價值的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"另外我們還可以幫助填寫病例，叫病例助手。醫生要花二三十分鐘寫一份病例，把各種觀察記錄上去。我們能不能根據從病例中學習的一些知識，幫助醫生來填。我們發現用知識庫、診療模型等可以幫助客戶省非常多時間。我們在一些大的三甲醫院做測試後，發現很多醫生都非常歡迎這樣的產品，因爲它幫醫生省了時間。這都是業務落地的場景。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們當然希望最後做的技術就是做一個病例輔導，隨着我們能夠收集更多數據，更多的醫院Deploy這些系統，就可以訓練出更好的模型，最後希望可以往診療方向走。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":"謝謝三位嘉賓對這個問題的分享。剛纔有兩個點大家都不同程度的提到，一個是怎麼樣能夠找到真正的痛點和需求，一個是怎麼樣能夠規模化落地。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我講一下我自己遇到的一個親身體會，AI，尤其是技術型的公司，往往會針對一個具體問題，定義得很清楚，去解決這個問題。但從產業或者企業的角度，往往是把整個解決方案提供出來。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"舉兩個物流行業實際的例子，一個是用光學識別的方法，快速識別物品要到達的地址。後來我們落地時發現這個數據實驗室很準，到場地不準。原來是上面有一個風扇在吹傳送帶，很薄的快遞就被吹跑了，解決的方法是把電風扇換個位置。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"另外一個例子，AGV在場地裏運輸貨品。一到實際場地裏，速度、精準度都有問題，主要是因爲灰塵太大。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"從企業的角度，你只給他解決一個技術上的單點問題可能不夠，他需要一個解決方案。我想知道，從氪信的角度，你們是會端到端把問題解決還是跟他們一起合作，把技術難點突破一下？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"   "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"朱明傑："},{"type":"text","text":"氪信是第一批成立CTO工作室的，主要做了些什麼呢？我們通過這段時間和幺寶剛老師合作，寶剛老師很多經驗，包括IDEA其他研究員豐富的經驗，對我們本身的工作和產品有很多啓發，這是很也共通的。大家面對的問題，不同行業都差不多。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"剛剛寶剛老師說在物流行業遇到什麼問題。在金融行業裏，我們說到底提供什麼東西時，一樣會遇到。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"這個事情是業務和技術之間的距離，或者叫人機距離。"},{"type":"text","text":"操作的人跟最後給他一個算法的東西，算法解決只是其中一個問題，這之間有很大的距離。業務員不懂到底怎麼回事，出來一個數字或者模型，要解釋給他。尤其AI又是黑盒子，在金融裏這個問題非常重，但在一開始時又不得不給他提供的是單點的技術問題。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"最開始時我們什麼都沒有，面對的客戶又是非常大的銀行。大金融機構被逼到內卷後沒有辦法，就要開始自動決策，會拋出一些單點的問題讓你來解決。AI公司都是去解決這樣的技術問題。但後面的客戶說他既不是那麼熟悉數據，對業務的瞭解也沒那麼多，還需要你去告訴他該怎麼做這件事情。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"我們提供的產品解決方案是AI部分要加上業務部分，所以我們把這個東西叫AI+Professional knowledge（專業知識）+Integration（融合），簡稱“API”戰略。"},{"type":"text","text":"我們過去幾年在給大客戶裏做的單點技術問題過程中，一直在學習業務的Know-How。把業務的Know-How變成內核是AI的技術框架，外面加一個業務解釋層，能讓業務的人懂，而且還能預製一些業務指導操作的經驗，這樣才能把產品給推廣出去。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"歸納下來，我們做這件事情，是需要在前期做單點，越到後面沉澱的東西越多，能固化下來的東西就越多。像最開始，大家拿的AI是電動機，不知道能幹什麼，我們就去不斷向主人科普：你洗衣服怎麼做的，我是不是可以幫你做一個殼？那就做成洗衣機；或者你需要做一個製冷的東西，我們可以做一個電冰箱。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"客戶要的是電冰箱、電飯鍋、空調、洗衣機，不是要你的電動機。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":"容力博士，自動駕駛現在可以說是風口，大家都在往這個方向投入、努力。肯定這裏面有很多需求，因爲它既是經濟問題也是社會問題，你們爲什麼選擇了重卡領域而不是乘用車領域？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"容力："},{"type":"text","text":"這裏一個關鍵點是重卡作爲運輸的，按照我們以前學的理論，它是一個生產資料或者生產工具。它跟我們平常用的車是不一樣的，作爲一個生產工具，你如果給它一點降本增效的效應，就會讓資本家有極大的興趣。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"卡車做自動駕駛和乘用車做自動駕駛都有一個共同點，現在技術不是完全成熟。如果在路上碰見一輛車在公開道路上開，車裏沒有司機，這還是比較罕見的，只有在非常少的限定的小區域裏測試可以做到。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"這個技術還沒有那麼成熟，但我們認爲重卡的自動駕駛將會率先落地，率先落地的邏輯是因爲我在能夠完全去掉駕駛員實現之前的情況下，就帶來降本增效的效益。因爲現在通過人工智能、大數據、算法，能夠讓人工智能的輔助駕駛在開卡車時比老司機還省油。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"大家做過物流行業，尤其是重卡駕駛，就知道我要是做一個物流企業尤其做一個車隊，一年的費用裏，將近25-30%是油費。如果通過人工智能自動駕駛方法，比老司機能省大概10%的油費。這是巨大的效益，尤其在物流行業內卷比較嚴重的情況下。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在完全無人駕駛實現之前，我們目前看到中的這個賽道，用重卡做無人駕駛，今年、明年就能達到這個效益。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們公司做現在的自動駕駛卡車，是前裝量產，跟一汽解放合作的，今年已經量產下線了。我們認爲我們過去兩年多的努力沒有白費，現在能把自動駕駛、人工智能技術落地到重卡上。如果我們去做乘用車，它是看起來很性感的東西，比如我們做人工智能做到了比特斯拉還強，這一定是新聞頭條"},{"type":"text","marks":[{"type":"strong"}],"text":"。但在我們作爲公司來看，我們看中的不是上新聞頭條，也不是把自動駕駛這個技術讓大家都知道。我們是想掙錢的，這也是爲什麼選重卡自動駕駛。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":"容力博士，你既然提到特斯拉，我必須問一句。特斯拉花了很多精力投入到FSD（完全自動駕駛）技術，在新聞上也做了很多這方面的宣傳。乘用車的FSD技術跟重卡有什麼關鍵的核心上的不同？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"容力："},{"type":"text","text":"技術本身差別不大，兩個流派：Waymo流派、特斯拉流派的。Waymo從收集數據開始做，大概現在有600臺車在路上跑。還有一個是特斯拉這種方式，一邊造火箭，一邊造車。火箭是星辰大海，車是現在就掙錢，車給大家帶來推背感的體驗。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在我看來和重卡相比，技術沒有太大區別。我們認爲特斯拉的路線可能是正確的，首先要讓車在路上跑，跑了纔有足夠的數據，有100萬輛車的自動駕駛在路上跑，最後才能說那麼快能實現FSD。這樣的數量規模是Waymo這幾百臺車沒法比的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"可以說我過去受到寶剛和其他同事們的教育，一直認爲大數據，100萬輛車的規模在路上跑，收集的各種各樣數據，這是無法用一個實驗室裏拍幾百臺車的數據能解決的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":"霖濤博士，剛剛您也介紹了你們在醫療領域的一些具體應用。但在我心目中翼方健數首先是一個很大的平臺公司，隱私計算也是你們很關注的領域。 回到需求，有一個說法，說隱私計算是“AI落地的最後一公里”，您同不同意這個說法？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"張霖濤："},{"type":"text","text":"我當然非常同意。我剛纔說到很多AI應用問題，我們不是傳統意義上的AI公司，是隱私計算公司，做數據和平臺的。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"問題在於我爲什麼要去做AI，隱私計算本身不一定有任何價值，價值最後還是要數據的價值，用戶覺得我需要這個東西。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"像朱博士剛纔談的電，我們之所以要做這些應用，就相當於先弄個燈泡。我怎麼能告訴你說電有用，這很難。你告訴我電有用，將來怎麼樣怎麼樣，人家說我現在有什麼用？你說我現在就能給你點亮燈泡。我們現在講的輔助診療是做這樣的事情。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們不希望說以後你家裏的電器都我包了，這不是我們的目的。"},{"type":"text","marks":[{"type":"strong"}],"text":"我們希望電給你布好了，將來有更加專業的公司能夠把其他事情做起來，"},{"type":"text","text":"比如醫院裏有各種信息化，比如藥企就願意用醫院的數據做藥物研發。比如保險公司會願意用醫院的數據對醫療的保險做很多控費，能夠更有選擇的進行保單的定價等等。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"像這樣事情，不一定是我們能做，當然我們也會去聯繫數據員和數據提供方。但我們也深深知道，數據全部用起來，數據的價值，這是幾十萬億的市場，肯定有很多很多公司在這個ecosystem裏大家一起貢獻。但最重要的事情是一開始能把這個數據先變成一個可用的情況。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"所謂可用的狀態，醫院裏以前信息化是各種公司的設備宕出來的數據，每一家醫院的影像設備都不是同一家公司造的。他們也無所謂，因爲這個數據是不準備給別人用的，只是萬一出了什麼事需要查，就拿來用。我怎麼把這個數據弄成將來被藥企、保險公司用。我們需要做這個數據的治理、清理、歸一和數據標準的制定。這些事情我們都得做，而這個東西也要用很多很多AI的技能，因爲不可能全部都用手做，我們在這個過程中用了很多工具。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"這麼多工具，難道只是爲了醫療上給大家填病例嗎？不是的。後面希望把這個事情skill起來，需要各個其他公司的貢獻或者一起協作。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":"謝謝霖濤博士。三位嘉賓都是有科學家頭腦的創業者，從企業角度，企業對投入產出還是很在意的。在AI落地過程中，很多時候遇到的問題是項目制和規模化。怎麼樣避免被困於定製化的瓶頸中？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"張霖濤："},{"type":"text","text":"也不是，這是所有To B、To G業務的痛點，我們公司主要是To B和To G，不像To C的業務，做一份東西所有人都能用，To B和To G往往是要定製的。我們也是深刻認識到這一點，其實沒有什麼特別好的答案。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們在過程中不斷地優化自己的能力。比如像做數據清洗，第一個醫院進去，折騰得不得了，爲了折騰這些工具，就建了很多自己的工具，用Machine learning的方法做得更快一點。到第二家醫院就發現好多了，但又出現各種各樣的問題，我們又得修。等做到第十家醫院，該見過的事情都見過了，希望是這樣做。從我們過去的經驗來說，確實是這樣。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"很多To C的公司也會遇到這樣的問題。第一個客戶和第二個客戶可能不一樣，但第1000個客戶和前面999個客戶已經覆蓋得差不太多了。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"   "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"容力："},{"type":"text","text":"對於自動駕駛來講，第一步就想得非常清楚我們要做一個產品，所以非常仔細的花一些時間定義這個產品，包括跟一汽解放合資成立像智途公司來發布這個產品，包括和我們的大股東滿幫集團，我們很早就定義了這個產品，按產品的方式做。這樣很早就規避了做項目，從第一個做到第十個的情況，取決於產業鏈最開始的建立，繞過了很多問題。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"這是我們公司過去走的路。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"朱明傑："},{"type":"text","text":" 解決這個問題，最重要是產品化。我們這幾年的實踐下來有兩點心得：一是用AI工程化的能力解決這個問題。我直接拿一個電動機給人，他就問你這個電動機能不能抽水，能不能幹啥，你每一個都要給他做一套。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們的AI決策引擎叫“非或然引擎”，外面套了一個Pilot“業務專家駕駛艙”的殼，縮短人機交互的距離，能讓業務人員更透明的看到輸出結果和業務之間的關係。這需要跟業務場景磨鍊很長時間，我們希望快速的做出“洗衣機”和標準的東西。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"再者，"},{"type":"text","marks":[{"type":"strong"}],"text":"就是跟大的頭部客戶做標杆，跟相對小的客戶做規模化拓展。"},{"type":"text","text":"因爲相對小的客戶自己知道他的業務知識也需要依賴於你，你的Pilot系統加“非或然引擎”給他的東西能夠解決他90%的問題，也會更信賴你。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"總結起來，還是響應沈向洋老師的建議：躬身入局。"},{"type":"text","text":"我們做了好幾年這樣的工作，從這兩年開始能看到一些收穫，逐漸把這個業務和形成的“非或然引擎”+Pilot系統集成API，在通用場景上把它推廣開。我們也希望以更快的速度讓更多客戶能夠用上這些東西。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":"我們剛剛都談了一些非常腳踏實地的話題。最後我想問一個仰望星空一點的問題：有一個說法是未來所有的企業都是AI企業。我也想請幾位暢想一下，我們離這個未來還有多遠？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"朱明傑：行業進入到內卷時代後，就不得不成爲AI企業。"},{"type":"text","text":"說白了就是人口紅利沒了，必須藉助工程師紅利，如果工程師紅利也不能解決，可能就留給未來去解決的問題吧。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"比如我就自己開一個包子鋪，不用什麼AI，只要把我家門口的生意做好就行了。但我要開一個連鎖包子鋪，要管理上千家企業，這麼多流程，一定要從淺層數字化到深層數字化做到，做到這件事，必然倒逼AI的企業轉型。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"容力："},{"type":"text","text":"說所有的公司都是AI的公司，看怎麼去理解。我認爲這是一個非常非常美好的前景。對於我們在座的大家應該是非常好的，一小部分公司會成爲AI的公司，其他的公司不得不借助我們在座的人，這些AI公司，去擁抱未來。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"爲什麼只有一小部分人才能做到真正的AI公司，其他的公司去仰仗我們這些公司做事呢？"},{"type":"text","marks":[{"type":"strong"}],"text":"因爲成爲一個AI公司並不那麼容易，要有AI的人才、文化，還要有AI的胸懷。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"張霖濤"},{"type":"text","text":"：我們認爲不一定所有的公司最後都是AI公司，但所有公司可能都是數據驅動的公司或者數據公司。這個數據要拿出來用，需要更多的expertise。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"比如要做AI，我們現在知道算法、算力、數據。不是所有的公司都有這三個能力這三個能力是非常非常少的combination。阿里巴巴有，但隨便一個社區醫院不可能有這方面的能力。我們隱私計算做的事情是希望把這個東西decompose。decompose的意思是提供數據的人不一定懂算法，做算法的人不一定有數據。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"像剛纔說的包子鋪概念，我開一家包子鋪，有很多數據，我的客戶都是誰，每天來我這兒喫多少包子。但問題在於你這個包子鋪一共就三個員工，不可能有這樣的能力。但他有數據，如果能把這個數據expose給別人，讓別人來用，並且在很小的摩擦，建立數據和服務，很容易做起來，這就是隱私計算希望做的事情。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"做成這件事後可能就有別人來爲他服務，有一個標準的應用，把他的數據分析，說你應該這麼調你的包子餡，因爲別人都是怎麼做的，你這個怎麼做的。"},{"type":"text","marks":[{"type":"strong"}],"text":"將來可能所有公司都是數據驅動或者AI的公司。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"   "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"幺寶剛："},{"type":"text","text":"這個話題是非常大的話題，隨着技術的發展，人才的集聚，外面的大環境包括人口紅利逐漸消失，產業升級的需求，這個問題也變得越來越重要，我們只是在這個領域從不同行業、不同方法去嘗試。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"今天非常非常感謝三位嘉賓的分享，也謝謝各位聽衆的時間，謝謝大家。"}]}]}