金融领域人工智能早期应用者的经验教训

2Finance transformation leaders (FTLs) pursuing AI are quickly realizing that AI adoptionwill take more time than anticipated. FTLs can read this research to learn from theexperience of early adopters of finance AI, better plan their AI adoption strategyand ensure a more effective rollout.OverviewKey findings•Despite being the first to use AI, early adopters of finance AI found that thespeed to implement AI was generally slower than they initially anticipated.•Talent and collaboration with enterprisewide AI efforts are key differentiatorsof early adopters of finance AI; early adopters more frequently had data scienceskills available to them and were more frequently involved in enterprise AIgovernance efforts.•Early adopters of finance AI found that they did not have to set their ambitionsand governance approach at the onset of their AI initiatives. Instead, they wereable to figure out those elements of their AI strategy as they experimented withthe technology.•The evaluation of ROI for AI remains a challenge, even for early adopters offinance AI. 3Data insightsOver the last year, finance has made significant progress in AI adoption. The percentageof finance functions using AI (i.e., with ongoing AI pilots, AI in production or AI being usedat scale) has gone up, from 37% in 2023 to 58% in 2024. This jump over the course of justone year is remarkable. In 2023, finance was a laggard in AI adoption compared to otheradministrative functions like HR, legal, real estate and procurement. But in 2024, financehas closed that gap and has done so due to significant efforts and investments fromacross the function.Indeed, the increase in AI adoption has been hard-earned. Almost half of finance leaderscurrently using AI report that their AI adoption went slower than anticipated. Only 11% reportthat the adoption went faster than anticipated (see Figure 1).Figure 1: Speed of AI adoption for early AI adopters in financePercentage of respondentsn = 70 finance leadersSource: 2024 Gartner AI in Finance Survey46%Slower thanexpected 43%As expected11%Faster than expected 4To avoid getting bogged down during the course of AI adoption, FTLs should take the timeto learn from the early adopters of AI in finance before charting out their own course. Basedon our 2024 Gartner AI in Finance Survey, Gartner has found that:•Having the right talent and enterprise-level involvement on AI governance andmanagement initiatives are key drivers of AI adoption success•Certain elements of AI strategy, such as the overall ambition the function hasaround AI and its full governance approach, can be determined as the organizationexperiments and learns more about the technology•Specific challenges remain, even for early AI adopters. In particular, financeleaders across the board are struggling to evaluate the ROI of AI solutionsThese findings are explored further in the following analysis.Talent and enterprise-level involvement are keydrivers of successOur survey shows two key differentiators between finance AI users and nonusers.Differentiator No. 1: Use of data scientist talentEarly AI adopters are over twice as likely to use data scientists within their financefunction compared to non-AI adopters (see Figure 2). These data scientists usetheir analytics and data management expertise to build and test AI models, preparedata for AI consumption and assist in the technical implementation of AI software.Figure 2: Finance functions using data scientistsPercentage of respondentsn variesSource: 2024 Gartner AI in Finance SurveyAI adopters (n = 70)0%35%70%Non-AI adopters (n = 49)61%27% 5The substantially higher rate of data science use among early AI adopters emphasizesthe importance of having the right type of talent within the finance function to helpexecute AI initiatives. However, this does not necessarily mean that finance leadersneed to hire for new data science roles in the finance function to get started. Only33% of early AI adopters hired data science talent from external sources. The resteither trained citizen data science talent internally (51%), leveraged data sciencetalent from other parts of the organization (44%) or engaged third-party providers(30%). See Figure 3.Figure 3: Placement of data scientists among finance AI adoptersPercentage of respondents (multiple responses allowed)n = 70 finance leadersSource: 2024 Gartner AI in Finance SurveyBPO = business process outsourcingInternal to finance(trained internally)Internal to company,but external to financeInternal to finance(hired externally)External (third party;e.g., consultants, BPO,independent)0%30%51%44%33%30% 6Differentiator No. 2: Involvement in enterprise-level AI effortsThe survey also suggests that enterprise-level involvement might be a key jumping-offpoint for finance function AI adoption. Indeed, 60% of finance AI users are engaged inenterprise-level AI governance efforts, compared to just 16% of finance AI nonusers.This data is echoed in the conversations our research team has