《AI調研分析:AI浪潮3.0——從運營到戰略.pdf》由會員分享,可在線閱讀,更多相關《AI調研分析:AI浪潮3.0——從運營到戰略.pdf(40頁珍藏版)》請在三個皮匠報告上搜索。
1、 2023 Gartner,Inc.and/or its affiliates.All rights reserved.Gartner is a registered trademark of Gartner,Inc.and its affiliates.This presentation,including all supporting materials,is proprietary to Gartner,Inc.and/or its affiliates and is for the sole internal use of the intended recipients.Because
2、 this presentation may contain information that is confidential,proprietary or otherwise legally protected,it may not be further copied,distributed or publicly displayed without the express written permission of Gartner,Inc.or its affiliates.Mike Fang高級研究總監AIAI調研分析:AIAI浪潮3.03.0從運營到戰略1 2023 Gartner,I
3、nc.and/or its affiliates.All rights reserved.RESTRICTED杯子是半滿還是半空?2 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED6 6個月前CIO們堅信:AI能提供巨大的業務價值(自動化,效率,客戶滿意度,營收,成本削減,環境、社會和治理,多元、平等與包容,)董事會成員認為:AI受到了炒作,其風險大于價值(昂貴、脆弱、偏見、可解釋性、隱私、安全、概率)3 2023 Gartner,Inc.and/or its affiliates.All rights
4、 reserved.RESTRICTED在 ChatGPT ChatGPT 推出之后董事會成員認定:生成式AI能提供巨大的業務價值(自動化,效率,客戶滿意度,營收,成本削減,環境、社會和治理,多元、平等與包容,)CIO們焦慮:生成式AI受到了炒作,其風險大于價值(昂貴、脆弱、錯覺、偏見、可解釋性、隱私、安全、概率)4 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTEDAIAI浪潮3.03.0正在襲來戰略運營重點:概念驗證差異化價值AIAI工程AIAI原生自主自適應生成式原生學習集成靈活透明以決策為中心韌性可
5、擴展大部分中國企業所處階段5 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED守成還是創新?6 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED暫時中止減少投資無變化增加投資不適用(如供應商或投資者)來源:放眼炒作之外:ChatGPT和生成式Al對企業的影響之網絡研討會民意調查,2023年3月30日和4月21日ChatGPTChatGPT問世以來AlAl投資的變化情況問題:自ChatGPT上市以來,您最近的AI投資策
6、略發生了哪些變化?占受訪者的比例7 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED如何規劃新的投資?守成?創新?8 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED全球70%的企業部署了20個以上的AI用例,全球企業比小型企業平均部署的AI用例更多當前部署的AI用例總數41個AI用例的平均部署數量2%28%43%21%6%0%40%80%5個以內520個2050個50100個100個及以上70%n=618,指涉及A
7、I的領導者,不包括回答“不確定”的受訪者問題06:貴公司目前部署了多少個不同的AI用例?來源:2022年Gartner AI用例投資回報率調研9 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED全球企業AI用例的平均使用壽命為3.5年AI用例的平均使用壽命3.5年AI用例的平均使用壽命n=618,指涉及AI的領導者,不包括回答“不確定”的受訪者問題07:貴公司AI用例的平均使用壽命是多久?來源:2022年Gartner AI用例投資回報率調研10%41%33%14%2%0%40%80%1年以內13年35年
8、510年10年及以上10 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED14%53%24%6%3%0%40%80%5個以內520個2050個50100個100個及以上中國86%86%的企業部署了5 5個以上的AIAI用例當前部署的AIAI用例總數2424個AIAI用例的平均部署數量86%86%n=300,指涉及AI的領導者問題06:貴公司目前部署了多少個不同的AI用例?來源:2022年Gartner中國AI用例調研11 2023 Gartner,Inc.and/or its affiliates.All
9、 rights reserved.RESTRICTED中國企業AIAI用例的平均使用壽命為2.82.8年AIAI用例的平均使用壽命2.82.8年AIAI用例的平均使用壽命6%63%26%4%1%0%50%100%1年以內13年35年510年10年及以上n=300,指涉及AI的領導者問題07:貴公司AI用例的平均使用壽命是多久?來源:2022年Gartner中國AI用例調研12 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED您可能認為需要預測性維護的工作AIAI13 2023 Gartner,Inc.an
10、d/or its affiliates.All rights reserved.RESTRICTED在設備上安裝傳感器連接傳感器與決策系統收集設備上的數據就數據意義達成一致分析歷史數據分析流數據監測KPI確定維護目標盤點資產和零件設定KPI在維護系統中集成AI規范物理環境準備數據運營化AI管理庫存/零件管理設施培訓維護人員管理優先事項和資源審查和改善提高AI信任度改變日程安排審查和改善測試維護系統AI需要預測性維護的工作14 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED0%1%9%20%48%20%2
11、%0%20%40%60%80%1234567使用AIAI后預期業務目標的完成情況評分:1=未完成任何業務目標;7=充分完成了所有業務目標中國五分之一的企業表示,其AIAI用例幾乎完成了所有業務目標未完成任何業務目標充分完成了所有業務目標n=274,指涉及AI的領導者,不包括回答“不確定”的受訪者問題03:貴公司迄今為止部署的AI用例,總體上來說預期業務目標的完成情況如何?來源:2022年Gartner中國AI用例調研4.84.8分業務目標完成情況平均分15 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED
12、2%5%22%54%16%0%30%60%1234567n=620,指涉及AI的領導者,不包括回答“不確定”的受訪者問題05:貴公司迄今為止部署的AI用例,總體上來說預期業務目標的完成情況如何?來源:2022年Gartner AI用例投資回報率調研業務目標完成情況評分:1=未完成任何業務目標;7=充分完成了所有業務目標全球70%的企業表示,其AI用例幾乎完成了所有預期的業務目標充分完成了所有業務目標未完成任何業務目標5.85.8分業務目標完成情況平均分70%70%16 2023 Gartner,Inc.and/or its affiliates.All rights reserved.REST
13、RICTED技術采用生命周期阻力人才技術治理客戶財務組織時間采用推廣階段試驗1-2+3產品產品組合業務業務業務轉型指標重點產品產品產品團隊團隊團隊來源:Gartner從AIAI項目到AIAI產品17 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED產品化運營,通過價值等式持續提升AIAI能力所提供的業務價值人工智能模型生產系統解決方案業務流程AI能力的價值業務價值的多個驅動力示例能力:通過AI來降低客戶流失率該模型在預測哪些客戶會流失和不會流失方面的準確度用于獲取模型預測的基礎設施的延遲和停機時間 要分類
14、的模型的客戶數量 留住客戶的成本被錯誤地預測為流失 正確預測客戶流失的留住成本 正確預測客戶流失的終身價值 客戶流失模型在生產中產生的價值+來源:改寫自Georgia Pacifica 例如,業務流程、業務定義或業務條件的變化可能會使模型的預測與現實脫節,從而導致模型的預測性能下降。模型的預測性能只是影響模型在生產中產生的商業價值的眾多因素之一。根據R2、RMSE或F1 Score等模型性能指標衡量,即使模型完全正常工作,業務價值也可能在生產中消散。a18 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED
15、價值還是風險?19 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED你是否認為生成式AIAI利大于弊?全球是75%否14%不知道11%中國不知道是否n=253n=253,指20232023年4 4月的網絡研討會參與者來源:放眼炒作之外:ChatGPT和生成式Al對企業的影響之網絡研討會民意調查,2023年4月21日20 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED管理好AIAI所帶來的風險價值風險21 2023
16、Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED所有模型都是錯的,但部分模型是有用的。英國統計學家George E.P.Box22 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED部署AI用例時,風險因素至關重要52%考慮風險,但風險不是部署AI用例的決定性因素36%開始不考慮任何風險,只考慮潛在的業務益處12%n=622,指涉及AI的領導者,不包括回答“不確定”的受訪問題11:部署AI用例時,風險因素在決策中的占比如何?來源
17、:2022年Gartner AI用例投資回報率調研部署AI用例時,風險因素在決策中的占比全球多數企業表示,部署AI用例時風險因素至關重要RESTRICTED23 2023 Gartner,Inc.and/or its affiliates.All rights reserved.中國部署AIAI超過4 4年的C C級高管和企業較少考慮風險因素部署AI用例時,風險因素在決策中的占比45%49%7%0%50%100%考慮風險,風險因素至關重要考慮風險,但風險不是決定性因素不考慮任何風險,只考慮益處C級高管(22%)明顯高于董事(2%)和經理(2%)采用AI超過4年的企業(13%)明顯高于采用2-4
18、年(4%)和不到2年的企業(3%)n=300,為涉及AI的領導者問題11:部署AI用例時,風險因素在決策中的占比如何?來源:2022年Gartner中國AI用例調研腳注ID:24 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED全球AI實施成熟度的驅動因素24 2023 Gartner,Inc.及/或其關聯公司版權所有。Gartner是Gartner,Inc及其關聯公司的注冊商標。法務專家參與AI用例的構思階段3.8倍對AI成熟度的影響n=638,指全部受訪者,不包括回答“不知道/不能說”的受訪者調研結果
19、基于邏輯回歸模型。*在90%時具有統計學意義,請謹慎解釋說明。問題07:現在或將來,AI在貴公司中的應用廣泛程度如何?來源:2021年Gartner P-21023企業中的AI調研3.8倍對AIAI成熟度的影響25 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED高業務價值投資管理削減機遇低業務價值低風險高風險來源:Gartner風險、價值和成本的處理象限26 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED技術創新
20、生命周期想法沙盒生產合成數據賦能的數據能力真實數據合成數據概念驗證減少了不可行計劃的數量合成數據有助于富達國際在沙盒中針對模型、供應商和實驗開展概念驗證數量:在真實數據不足的情況下增加數據數量和種類。數量有限數量更多噪音:測試模型對“噪音”(如黑天鵝事件)的反應。普通事件稀有事件平衡:平衡數據集,以便有效訓練模型。效用低效用高可共享性:數據完全匿名,不僅是數據脫敏,以便與合作伙伴共享數據。脫敏數據匿名數據可用性:部分實驗可能需要真實/歷史數據集不存在的數據能力。來源:根據富達國際案例編寫數據利用隱私計算技術管理數據泄露風險,充分挖掘數據價值27 2023 Gartner,Inc.and/or
21、its affiliates.All rights reserved.RESTRICTED人才還是組織?28 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED如何根據組織實際情況選擇到合適的人才?組織?人才?29 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED中國人才或服務提供商短缺是AI采用的最大障礙AIAI采用的最大障礙53%難以找到優秀的人才、軟件供應商或服務提供商30%難以在企業中確定有價值的AI用例14%
22、難以理解AI技術在用例中的益處3%其他難以選擇和訓練AI模型難以找到同時擅長業務和AI技術的人才n=294,指涉及AI的領導者,不包括回答“不確定”的受訪者問題15:在貴公司中,AI采用的最大障礙是什么?來源:2022年Gartner中國AI用例調研成本高昂;業務評估指標不明確30 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED全球超過70%的企業表示,AI人才并不是難點AI人才需求擁有/能夠獲取人才72%人才短缺28%19%26%27%19%10%0%50%100%我們已經擁有充足的人才我們可以通過培
23、訓已有的內部員工來滿足相應的需求我們可以獲取到所需的人才我們略微缺乏人才我們長期缺乏人才n=698,受訪者基數:不包括回答“不確定”的受訪者問題15:您如何看待貴公司AI發展需要的人才或技能情況與現有的人才或技能情況?來源:2021年Gartner P-21023企業中的AI調研31 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTEDITIT技能領域理解AI技能數據科學家計算語言學家知識工程師流程工程師運籌學專家數據工程師系統架構師主題專家業務轉化師應用開發人員業務線經理企業策略機器學習工程師機器學習架構師
24、不同AIAI角色在組織中的重要性32 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED全球治理結構決策權中心化vs.去中心化?32 2023 Gartner,Inc.及/或其關聯公司版權所有。Gartner是Gartner,Inc及其關聯公司的注冊商標?;旌辖M織模式中心化AI團隊使用分布在企業內部的AI特別小組1.6倍對AIAI成熟度的影響n=638,指全部受訪者,不包括回答“不知道/不能說”的受訪者調研結果基于邏輯回歸模型。*在90%時具有統計學意義,請謹慎解釋說明。問題07:現在或將來,AI在貴公司中
25、的應用廣泛程度如何?來源:2021年Gartner P-21023企業中的AI調研33 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED廣泛的數據分析經驗數據分析專業知識業務和數據領域知識0月18月30月成熟階段初始團隊基礎數據科學高級數據科學123交付成果 數據獲取 數據質量 報告 通用式ML 預測分析試驗 生產模型 專門的任務來源:Gartner基于組織的人工智能人才戰略34 2023 Gartner,Inc.and/or its affiliates.All rights reserved.REST
26、RICTED無論從哪個角度觀看,AIAI都已經成為企業必須擁有的核心能力守成為先管理風險擇人任勢總結:建議我們的CIO們結合以下三點:來更加戰略地計劃,應用,運營AI。35 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTEDAccess to Gartner research is subject to entitlement.For information,please contact your Gartner representative.GartnerGartner相關報告推薦Chinese AI S
27、urvey Analysis:AI Trends Wave 3.0 From Operational to StrategicCase Study:AI Model Operations at Scale(Fidelity)Case Study:Skills-Adjacent Data Science Team Development(Eastman)Case Study:Monitoring the Business Value of AI Models in Production(Georgia Pacific)36 2023 Gartner,Inc.and/or its affiliat
28、es.All rights reserved.RESTRICTEDAccess to Gartner research is subject to entitlement.For information,please contact your Gartner representative.謝謝!37 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED2022 Gartner China AI Use-Case Survey:This survey was conducted to understand AI
29、 implementations in China,and to understand where organizations have been most successful in deploying AI use cases.The research was conducted online from 14 November through 16 December 2022 among 300 respondents from organizations in China.Quotas were established for company sizes(in terms of annu
30、al revenue)and industries to ensure a good representation across the sample.Quotas included 45 small businesses(less than$50 million),105 midsize enterprises($50 million to less than$500 million),120 large enterprises($500 million to less than$10 billion),and 30 global enterprises(over$10 billion).O
31、rganizations were required to have developed AI to participate.Respondents were required to be in a manager role or above and have a high level of involvement with at least one stage of the life cycle from ideating to measuring AI use cases.Disclaimer:The results of this survey do not represent glob
32、al findings or the market as a whole,but reflect the sentiments of the respondents and companies surveyed.Appendix-Methodology38 2023 Gartner,Inc.and/or its affiliates.All rights reserved.RESTRICTED2022 Gartner AI Use-Case ROI Survey:This survey sought to understand where organizations have been mos
33、t successful in deploying AI use cases and figure out the most efficient indicators that they have established to measure those successes.The research was conducted online from 31 October through 19 December 2022 among 622 respondents from organizations in the U.S.(n=304),France(n=113),the U.K.(n=10
34、6)and Germany(n=99).Quotas were established for company sizes and for industries to ensure a good representation across the sample.Organizations were required to have developed AI to participate.Respondents were required to be in a manager role or above and have a high level of involvement with the
35、measuring stage and at least one stage of the life cycle from ideating to testing AI use cases.Disclaimer:The results of this survey do not represent global findings or the market as a whole,but reflect the sentiments of the respondents and companies surveyed.Appendix-Methodology39 2023 Gartner,Inc.
36、and/or its affiliates.All rights reserved.RESTRICTED2021 Gartner AI in Organizations Survey:This study was conducted to understand the keys to successful AI implementations and the barriers to the operationalization of AI.The research was conducted online from October through December 2021 among 699
37、 respondents from organizations in the U.S.,Germany and the U.K.Quotas were established for company size and for industries to ensure a good representation across the sample.Organizations were required to have developed AI or intended to deploy AI within the next three years.Respondents were require
38、d to be part of the organizations corporate leadership or report into corporate leadership roles,and have a high level of involvement with at least one AI initiative.Respondents were also required to have one of the following roles when related to AI in their organizations.determine AI business obje
39、ctives,measure the value derived from AI initiatives or manage AI initiatives development and implementation.The survey was developed collaboratively by a team of Gartner analysts and Gartners Research Data,Analytics and Tools team.Definition:Artificial Intelligence applies advanced analysis and logic-based techniques,including machine learning,to interpret events,support and automate decisions,and to take actions.Disclaimer:Results of this survey do not represent global findings or the market as a whole,but reflect the sentiments of the respondents and companies surveyed.Appendix-Methodology