2024年智能炼金术:生成式人工智能如何彻底变革现代企业中的商业智能和分析白皮书_第1页
2024年智能炼金术:生成式人工智能如何彻底变革现代企业中的商业智能和分析白皮书_第2页
2024年智能炼金术:生成式人工智能如何彻底变革现代企业中的商业智能和分析白皮书_第3页
2024年智能炼金术:生成式人工智能如何彻底变革现代企业中的商业智能和分析白皮书_第4页
2024年智能炼金术:生成式人工智能如何彻底变革现代企业中的商业智能和分析白皮书_第5页
已阅读5页,还剩54页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

TheAlchemyofIntelligence:

HowGenerativeAIcan

revolutionizeBusiness

IntelligenceandAnalyticsinModernEnterprises

TableofContent

Introduction 03

BusinessUser 04

Opportunities 04

Challenges 04

Recommendations 07

BusinessAnalyst 08

Opportunities 08

Productivity 08

ProgrammingforNon-Programmers 10

Insights 10

Beautification 11

Challenges 12

Usefulness 12

Trust 13

HumanErrorandDocumentation 14

Security 14

Recommendations 15

Test 15

Adopt 16

Train 16

DataAnalyst/CitizenDataScientist 17

Dylan’sTransformation 18

ArrivalofAIAgent 19

ATeamofAgentsEmerges 21

Summary 21

ITAdministrator 22

InfrastructureDemands 22

DataGovernanceandSecurity 23

Observations 24

SystemArchitect 25

Opportunities 25

Challenges 26

Recommendations 26

Summary 27

Conclusion 27

Authors 28

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA2

Introduction

Intherapidlyevolvinglandscapeoftechnology,businessesare

constantlysearchingforinnovativewaystostayaheadofthe

curve.OnesuchgroundbreakingadvancementisGenerativeAI,atechnologythathasthepotentialtoreshapethefutureofBusinessIntelligence(BI)andanalytics.Imagineaworldwheredataspeaksdirectlytoyou,whereyouranalyticstoolsnotonlyansweryour

queriesbutalsoanticipateyourneeds,providinginsightsyouhadn’tevenconsidered.ThisisthepromiseofGenerativeAI–atoolthattransformsrawdataintorich,actionableintelligence,empoweringbusinessestomakesmarter,fasterdecisions.

Thejourneythroughthiswhitepaperwilltakeyouintotheheartofthisrevolution.We’llexplorereal-worldscenarioswhere

GenerativeAIactsasacatalystforenhancedproductivity,sharperinsights,andmorebeautifuldatavisualizations.Frombusiness

userslikePeggySue,whoexperiencethemagicofAI-powered

chatbots,todatascientistslikeDylanDawson,wholeverage

generativemodelsforunprecedenteddataanalysis,thenarrativeunfoldstorevealbothopportunitiesandchallenges.Bytheendofthisexploration,youwillunderstandnotonlythetransformativepowerofGenerativeAIbutalsohowtoharnessiteffectively

withinyourenterprise.Forsimplicity,wehavebrokenthisintovariousreal-worldpersonas.

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA3

BusinessUser

UsesdashboardsandreportsgeneratedbyBItoolstomakeinformedstrategicdecisions.

Businessusers,likePeggySue,aretheworkerbeesofany

corporatehive.Checkingnumbershere.Doingtheworkthat

needstodobedonethere.Buzz.Buzz.Buzz.Thissectionexplorestheuniqueopportunities,challengesandrecommendationsfor

otherslikePeggySue.

Opportunities

PeggySuewasthrilledtohavethischancetolaunchhercareerwithaglobalbeerdistributioncompanyknownforthequalityofitsbeersandforbeingarealhigh-techleaderintheindustry.ShehadmanycoursesattheUniversitysheattendedondashboardsandanalytics,andtheyreallypaidoffforherduringherfirst6

months.Neverdidadaygobywhenshedidn’tseepostsonherLinkedInfromthisgoodfriend,orthatfriend,ravingabouttheirexperienceswithsometypeofgenerativeaichatbots.

Shewasthrilledthedayshereceivedanemailstatingthat

herorganizationwouldbegettingachatbotalongsidetheir

dashboards.Suddenlythereitwas,andPeggySue’sheartwasallaflutterwiththepossibilities.

Everythingshereadusedphraseslike“Gamechanging”“makeslifesomucheasier”“willreplaceallworkerseverywhere”somepostersmightaswellhaveusedthewords“hocuspocusdominocus”

becauseitsoundedlikemagic.

PeggySue’smindwasracing“LookatthebeautifulinputboxwhereitsaysIcanaskanything.”UnfortunatelyforPeggySueanother

thoughtstruck,“Icanaskanything,butIhavenoideawhattoask”

Challenges

Whilemanyorganizationsrushtogetabotintothehandsof

businessusers,blankcanvasparalysiscantakeoverbecausetheyfocusedonthetechnology,andnottrainingtheirstaffhowtouseit.

EventuallyPeggySuebeganaskingthequestionsastheycame

tohermind“TellmethetotalsalesforourbeerinSouthAmerica.”

“WhichlocationissellingthemostofourPorters?”“Whichdivisionisn’tdoingwellfinancially?”Eachofherquestionsreceivedananswer.

Thechallengeforherwasthatmostanswersjustseemedwrong.Whenshedugintothedetailedrecordsinherdashboard,she

confirmedtheywerewrong.“Well,Ireckonthisthingisn’tverygoodatmath.WhydidtheygivethisthingtomeifIcan’taskittoaddupnumbers?”

Othertimesthefigureswereaboutmeasuresthatsheknewthecompanyhadmultiplewaysofcalculating.“Thisanswermayberightforoneofthemeasures,buttheanswerdoesn’texplainwhichcalculationmethodisevenused.Evenifitsaccurateforonemethod,Ihavenowayofknowingforsureit’sthemethodmybossexpectstosee.”

“Maybeit’stheresoIcanaskquestionsaboutthedashboarditself”

shethoughttoherself.Whichwasgoodbecausealthoughshehadreceived10minutesoftrainingfromafranticallybusytrainer,shedidn’tremembereverything.So,sheasked“HowdoIfigureout

whichdivisionisstrugglingonmydashboard?”

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA4

Whilethoughtprovoking,shewashopingforspecificinformationaboutthedashboardshewaslookingat.Afterafewquestionslikethisshegotalittleworriedthatperhapsmanagementwastrackingherquestionsandthatifshekeptaskingquestionslikethis,she

mightbereprimandedforhavingnotalreadylearnedeverythingaboutthedashboardevenafterthewhopping10minutesof

trainingshehadreceived.

Onedayasshewasreviewingsomequarterlyfiguresandhercolleagueswereoutoftheoffice,somethingstruckher:

“MaybeIshouldbeaskingthesametypeofquestionsInormallyaskthem.”So,shedid:“WhataresomereasonsthatcouldexplainwhywearesellingsomuchmoreBrownAlethanotherbeers?”

Atthatmomentalightbulbwentoff,andachoruswassinginginPeggySue’shead.Assheproceededwithheranalysis,shewasagaincuriousaboutthedata.Althoughsalesweresohighfor

BrownAle,theprofitsweren’t.

Shequicklytyped“Whataresomereasonswearenotmaking

muchprofitonbrownaleconsideringwesellsomuchofit?”intothehandylittle“AskAnything”inputboxandwasagainimpressedwiththeresponse.

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA5

PeggySuewasinspiredbythispatternofaskingwhenshewas

puzzledaboutwhatcouldexplainthingsthatshedidn’tseeinthebarchartsandpiechartsandlinechartsonthescreen.Aftera

meetingonedaywheresheheardaboutacontestthecompanywashavingwhereanyemployeescouldmakesuggestions

abouthowtoincreasesalesshedecidedtogetreallyboldinherquestioning:“CanyoutellmeculturallywhywearesellingsomuchBrownAlewherewedoandwhatotherculturesaresimilarthatwecouldstartsellingitto?”

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA6

Recommendations

Whileonthecruiseshetookafterwinninghercompany’s

suggestioncontest,PeggySuehadmanychancestorecount

herexperiencestodatewithGenerativeAIinsidehercompany’sBusinessIntelligencetooltootherpassengers.

•Don’taskquestionsofanykindthatinvolvemath.

•Realanswerstorealbusinessproblemstypicallyinvolvecomplicatedbooleanlogicthatturnthemillionsofrows/columnsofdataintotruth,thatyourmodelmaynothaveaccessto.

•Don’taskforanswers,askforadvice.Answersimplyyouaredoneandwillact,butadviceimpliesyouwillaugmentthe

inputwithyourownknowledgethenact.

Onepassengershetalkedtooveroneofthosetalldrinkswith

fruitwedgesandanumbrellasaid“Wehave175differentBI

applicationsthatIworkwith.Whichoneofthemdoyouthinkis

therightonetostartusingwithoneofthoselargelanguagemodel

chatbotthingamajiggies?”PeggySuehadafewbitsofadviceforhim:

•TheoneusedbythegroupthatyouhaveprovidedsomeAILiteracytrainingtobeforehand.

•Theonethatyourbusinessuserspeekoverthecubiclewallsandchatwitheachotherthemostabout.

Storytellingasideforamoment...thebiggestrecommendationwecanofferforBusinessUsersistothinkofyourGenerativeAIchatbotslikeyouwouldanyothertrustedadvisorinyourlife.

•Theyaren’tgoingtodoyourworkforyou.

•Theywon’talwaysprovideadviceyouagreewith.

•Unlikeotheradvisorsinyourlife,theyarenevertoobusyforyoutoask,andtheynevergetoffendedwhenyouaskthe

samequestion10differentways.

•Youarestillultimatelyresponsibleforyourwork,soalwaysuseyourownintelligencetoaugmentanyadviceyoumayreceive.

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA7

BusinessAnalyst

WorkscloselywithstakeholderstounderstandbusinessrequirementsandusesBItoolstocreatereports,dashboards,andvisualizations.

Let’srewindtheclocksixmonthspriorandlookathowPeggy

Sue’snewBIcopilotcametobe.SallySue,theunstoppable

business-analyst-turned-datascientist,hasbeenexperimentingwithGenerativeAIforhercodingtasks.Copilotsareexcellentatgeneratingcodeandsummarizinglargeamountsoftext,andherbusinessrecentlyadoptedaBItoolthathasacopilotbuiltintoit.“Wow!”shethought.“Icananalyzemydataandbuilddashboardsjustbyaskingquestions?”Sallywasthrilledattheidea–aswas

herCIO.Canyouimaginethenumberofquestionsthatcouldbequicklyansweredifpeoplecouldchatwiththeirdataanddashboards?

Beyondtheexcitement,Sallyrealizedthatthereareseveral

potentialrisks.She’staskedwithevaluatingthiscopilotfor

productionandsendingitovertobusinessuserslikePeggy

Sue.WhatkindsofquestionsmightPeggyask?Whatkindof

dashboardswouldpeoplebuildwiththis?Howdowecertifythisforproductionuse?Whataboutdatasecurity?Isthereavariablecosttousethis?ThereareanumberofquestionsthatcametoSally’smind.Shebrokeherquestionsdownintotwomainareas:opportunitiesandchallenges.

Opportunities

GenerativeAIbringsampleopportunitiesforworkingwithdataanddashboardsbyhavingaconversationwithit.Sally’sgoingtofocusonthreeofthesepotentialopportunities:

1.Productivity-CanGenerativeAIimprovetheproductivityofbothmyjuniorandseniorbusinessanalystswhenworkingwithaBItool?

2.Insights-Canmystakeholders“chatwiththeirdashboard”togetfastertimetoinsight?

3.Beautification–CanGenerativeAIhelpcreatebetterlookingbeautifuldashboardswithbest-practicesautomatically

builtin?

Let’sexplorethesethreeconcepts.

Productivity

Buildingdashboardsisnoeasytask.Therearemanyconsiderationsthatmustbeaccountedfor:

•Who’stheaudience?Anexecutive?Abusinessunit?Anotheranalyst?Yourself?

•Whatmetricsdotheycareabout?

•Doesthedatasupportthosemetrics?

•Howoftenwilltheybeviewingthedashboard?

•Whatfollow-upquestionsdoyouanticipatethemasking?

•Doyouneedtosplitthisintomultipledashboards?

Theanswerstothesequestionswillgreatlychangethedesignofthedashboard.Understandingtheoverallbusinessproblemandhowthedatacansupportthosemetricsis,firstandforemost,

whatmustbedone.Forabrand-newbusinessanalyst,thisis

tough.Thismaymeansendingoutalotofemailstryingtogetanunderstandingofwhatmetricspeoplecareabout,wherethatdatalives,andwhatdocumentationtoread.

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA8

SallySue’sBIcopilotenableshertouploaddocumentationto

createagoverned,customcopilotcapableofansweringspecificquestionsmoreeffectively.Herorganization’sinternalwiki,whichhasimproveddocumentationpracticesovertime,servesasa

primaryresource.However,manyanswerstobusinessquestionsremainburiedindecade-oldemailsandtribalknowledge.

Duringhertestingphase,Sallytakesadvantageofthecontent

inthiswiki.Largelanguagemodelsexcelatparsingand

summarizingvastamountsoftext:themorequalityandaccuratedocumentationsheprovides,thebetterthecopilotwillperform.Sallypoursthroughthewiki,confirmsitsaccuracybycross-

checkingwithotherdepartments,scrapescurateddocumentation,carefullycleansit,andformatsitintoaJSONfilewithinformation

suchasthetitle,sectionname,andtext.ThisJSONisthen

uploadedintotheBItool’scopilotwherethesoftwarehandlestherest.

Customizingthecopilotwiththiscomprehensiveandaccurate

documentationenhancesitsabilitytoanswerbusinessquestions.Forexample,itcannowprovidedetailedexplanationsonlegacyprocessesthatareonlydocumentedinthewiki.Thisimprovementbooststheproductivityofjuniorandnewcolleagues,whousethecopilotforcreatingdashboardswithoutneedingtosiftthrough

extensivehistoricaldata.Notethatachievingthesebenefits

involvesongoingeffortstomaintainandupdatetheAImodeltoreflectcurrentbusinessneedsanddataqualitystandards.

Considerthisexample:

SallySuehasloadedinaknowledgebasethatdescribestheTTCKPI.Throughouttheoffice,everyonetalksaboutTTC–it’sthe

mostimportantmetricforpredictingprofitability.Butwhatdoes

TTCactuallymean?Intheknowledgebase,TTCisdefinedas“TotalTimetoCompletion.”Thelowerthisnumberis,thefasterbeer

canbottledanddistributed.ConsiderascenariowhereajuniorbusinessanalystistaskedwithcreatingadashboardaboutTTC.ThatjunioranalysthasthedatawhichholdsavariablecalledTTC.Theysimplyasktheircopilotthevaguequestion:

“Explaintomewhat’sinthedata.”Thecopilotthenreturns:

Yourdataappearstocontainanumberofbeerproductsproducedovertimeaswellasattributesaboutthoseproducts,suchas

weight,size,andcosttoproduce.TheTotalTimetoCompletion(TTC)isalsoincludedasoneoftheseattributesandisthe

business’smostimportantmetricbecauseitishighlynegatively

correlatedwithrevenue.ThelowerTTCis,themorerevenuetendstoincrease.Herearesomesuggestionsforvisualizingthedata:

Thekeythinghereisthatthecopilot:

•KnowswhatTTCisandcandecodewhattheacronymmeans

•KnowsthatTTCisimportantforrevenue

•Givesinitialvisualizationsuggestions

RatherthanrandomlyaddingvisualsintoaBItool,thecopilotis

givingtheuserastrongstartingpointandhelpsreducetheblankcanvaseffect.SallySueisalltoofamiliarwiththeblankcanvas

effect:it’sthefeelingyougetwhenyou’retaskedwithstartinga

brand-newpresentation,paper,ordashboard.You’represented

withablankcanvas,whichcanbeeitheragreatthingthatinspirescreativity,oraterrifyingthingthatsuccumbsyourbraintothe

dreadedwriter’sblock.AgoodBIcopilotcan,andshould,eliminatethiseffectandgivetheuseragoodstartingpoint.

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA9

StartingfromscratchcanbeadifficulttaskforeventhemostseasonedBIveterans

Somequestionstheusermightaskare:

•“GivemesomesuggestedvisualizationsforTTC.”

•“BuildmeastarterdashboardforaCEOwhocaresaboutrevenueasitrelatestoTTC.Includeothermetricsthatmaybeusefultoknow.”

•“Modifymydashboardsothatit’smoreaboutTTCovertimeratherthanTTCasawhole.”

SallySuetriesallthesequestionsandevaluateshowtheBItool

does.Ifit’swell-tuned,itshouldgivestrongstartingvisualizationsandmetrics.Shefindsthatitdoesanokayjobcreatingastarterdashboard.It’snotperfectandsomeoftheKPIsseemabitoff,

butit’scertainlynotbad,either.Thecopilotcoulddowithalittleimprovementfromuserfeedbackandadditionaldocumentation,butshe’llgettothatlater.Thevisualizationsitbuildsinitscurrentstateareatleastgoodforeditingandspurringnewideas–exactlywhatitshouldbedoing.

ProgrammingforNon-Programmers

Most,ifnotall,BItoolshavesomesortofprogrammingor

scriptinglanguagebuiltintothemsothatuniquemetricscanbe

createdonthefly.Thisiscrucialforcreatinghighlycustomized

dashboardsandgeneratingtheneededmetricsdirectlyinthe

toolwithoutthetedioustaskofleavingit,usinganothertool

orlanguage,thenreloadingthedata.SallySueiswell-versedin

programming,butherbusinessusersarenot–infact,she’sluckyiftheyknowSQL.Timeandtimeagainshegetsquestionsfromherusersonhowtocreatesomeofthemostbasiccalculations:True/Falseflags,summationsovertime,summationsbygroups,nestedcalculationsandmore.Sallynoticedthathercopilotincludesa

placetodescribecalculationstogeneratethem.Intrigued,shetriedasimpleprompt:

“AverageTTCbyregion.”

ThecopilotreturnsafewoptionsofaverageTTCgroupedby

region,allvariableswithinthedata.Thecodeitreturnsiswell-

formatted,commented,andevenincludesafewexamplevaluesforverification.Sallyisextremelyhappytoseethis,asitgivesherbusinessusersasignificantlyeasierwaytocreatemetricsand

customcalculations.Shesuspectsthatthiswillgreatlyreducetheamountofquestionsthatshegetsandimprovethespeedand

accuracyofdashboardcreation.

Insights

Picturesareworthathousandwords,andadashboardismade

ofmanyinteractivepictures.Peoplelovedashboardsbecause,

whendoneright,theycanproduceawealthofinformationina

compactspace.Ifyou’reabusyexecutive,youmighthaveaccesstodozensofdashboards.Somedashboardsarelargerthanothers,andsomerequireyoutoclicktoaspecificlocationandhighlight

specificpartsofthedashboardtogettheinsightsyouneed.

Sometimesyoudon’thavethetimenorthepatiencetogothroughthatdashboardtogetwhatyouneed.

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA10

Whatifyoucouldsimplyaskforit?

Whileplayingaroundwiththisnewtool,SallySuenoticedthat

hercopilotwasavailablenotjustfordashboardcreators,but

forreportviewersaswell.Shecouldaskquestionsaboutthe

dashboardtofigureoutwhatsheshouldreallyknowin30secondsfromthisseaofinformation.ShebuiltabasicTTCdashboardthatshewouldexpectexecutivestolookat.Itusedcommonmetrics

thatsheensuredwereapartofthedocumentationsheuploaded,andtherewasnocomplexmath.Sheincludedsomevisualslike

timeseriesgraphsofaverageTTCovertime,thecurrentaverageTTC,averageTTCfromthedayprior,andaverageTTCacross

variousregions.Sheaskedthecopilot:

“Givemethe3mostimportantthingsIshouldknowfromthisdashboard.”

Thecopilotreturned:

ThisdashboardappearstodisplaytheaverageTTCmetricovertimeandacrossregions.Thesearethemostimportantthingstoknowfromthisdashboard:

1.AverageTTCis3.5minutesandhasincreasedby30s(+20%)comparedtoyesterday.Thismayaffectrevenue.

2.AverageTTChasbeentrendingdownsinceJan2024,

buttherewasatemporaryspikeonMar23rdwhereTTCincreasedbynearly90s,butthenreturnedtonormalthefollowingday.

3.TheSoutheastregionhadthehighestincreaseinTTConMar23rd,contributingabout75%totheoverallriseinTTConthatday

SallySuethenasked,

“TellmethedriversofTTCfortheSoutheastregion.”Thecopilotreturned:

ThethreemostimportantdriversofTTCfortheSoutheastregionare:

•Bottlesavailable:thismetricappearstoberelatedtothe

numberofbottlesthatareavailableforproductiononadailybasis.Thisnumberrangesfrom0to10,000andispositivelyrelatedtoTTC.

•Totalworkers:thismetricindicatesthenumberofworkersonthewarehouseflooratthetimeandrangesfrom3to24.

•Unittemperature:thismetricindicatesthetemperatureofeachproductionunitandrangesfrom100Fto230F.Unitsover175Fareconsideredoverheating.

Sallywasimpressedwiththeperformanceofthecopilottograbinsightsfromthedashboard,showingthingsthatarebothdirectlyshownwithinthedashboardandthingsthatmaybehidden;

however,thisisjustfromherinitialtesting.Whileitcertainly

lookedconvincing,shestillneedstospendtimeverifyingtheaccuracyoftheseresultswhichsheplansondoinginafocusedtrustandsecuritytest.

Beautification

Sallyknowsalltoowellhoweasyitistodrag-and-droptobuild

dashboards.ModernBItoolsgenerallyhaveanoptimalsetof

colorsandsettingsturnedonforyoubydefault.Thesetendto

workwellandareusuallysetbyUXtoenablepeopletocreate

decent-lookingdashboardswithoutneedingtothinkasmuch

abouttherightcolorsorgraphsettings.WhatSallySuealsoknewishoweasyitistobuildbaddashboards.

Whatisabaddashboard?You’veprobablyencounteredone.Toomanymetrics.Numberseverywhere.Dozensofpages.Somanygraphscrammedintoasinglepagethanitbringsan8Kmonitortoitsknees.Colorsthatmakeyouwanttowatchblack-and-whitemoviesjusttorelaxyoureyeballs.You’vemostcertainlyseenabaddashboard.

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA11

Whatisevenhappeninghere?

Nobodygoesoutoftheirwaytobuildabaddashboard–likethatever-growingjunkdrawerinyourkitchen,itjusthappensover

time.Onenewmetrichere.Onenewgraphthere.OnenewpagefortheaccountantoverinBusinessUnit274.Itgrowsandgrows.Themoreeyesareonadashboard,themorelikelyitistogetthisway.GenerativeAIhasthepotentialtocurbthis.

AnygoodcopilotinaBItoolwillhavebeentrainedon

dashboardingbestpractices.AsSallywenttobuilddashboards,shepaidspecialattentiontothegraphsitcreated:

•Didtheymakesense?

•Arethecolorsappropriate?

•Aretheretitleswheretheyshouldbe?

•Diditcreatetheoptimalnumberofpages?

•Diditfollowbestpracticesformetricsonasinglepage?

•Isitaccessible?

AgoodBIcopilotfollowsdashboardingbestpracticesandgivesastrongstartingpoint

Thankfully,hercopilotfollowedallthesebestpracticeswhen

buildingadashboard,andevenhadtheabilitytogivesuggestionsonhowtoimproveexistingdashboards.Itseemsthatthe

designersofthiscopilotthoughtwellaboutthis.

Challenges

Overall,SallywashappywiththeBIcopilot’scapabilities.Hertestsweresimple,butsheneededawideraudiencetoreallytestitout.Assherolledoutteststoherotherbusinessanalysts,shehad

threeissuesinmind:

•Howusefulisthis?

•Canitbetrusted?

•Isitsecure?

Usefulness

ABIcopilotisanoptionalfeature,firstandforemost.Itsgoalistoassistyoutoexploreyourdataandbuilddashboardsfaster.

THEALCHEMYOFINTELLIGENCE:HOWGENERATIVEAICANREVOLUTIONIZEBUSINESSINTELLIGENCEANDANALYTICSINMODERNENTERPRISES|LFAI&DATA12

Sallyknewthat,likeanyotheroptionalfeature,itwillgocompletely

unusedifit’snotactuallyhelpful.Whensherolledoutthecopilottomorebusinessanalysts,sheaskedthemtopaycloseattentiontothefollowingquestion:doesthisfeaturemakebuilding

dashboardsfasterforyou,orisitafrustratinghindrance?

Ifyoufindyourselfgoingbacktothedrag-and-dropmethod,youprobablyfindthecopilottonotbeveryhelpful.Ifyouavoidthecopilotbecauseyoucan’ttrustitsanswers,thenit’snotagreat

copilot.Copilotsshouldbeconsistentintheiranswersandhavebestpracticesbuiltin.Ifit’screatinguselessdashboardsthat

aren’tevengoodforediting,thenthecopilothasfaileditsgoal.

Thecopilotshouldhelpreducetheblankcanvaseffect.Editingis,ingeneral,fasterthanstartingfromablankcanvas.Ifeditingis

harderthandragginganddropping,thenthecopilotisnotagoodfit.

Considercreatingasurveyorevenaworkshopforagroupof

users.Givethemsomesimpledatatoworkwithandaskthemtobuildadashboardoutofitusingthecopilotinalimitedamountoftime.Thedatashouldbeneutralandideallyhasnotbeen

seenbyanybody,butalsoeasytounderstand.Onegreatway

tofinddatalikethisistosearchforopendatasetsfrom

https://

.

Splitthegroupintotwo:onewhichhasaccesstothecopilot,andonewhichdoesn’t.Askthegroupwhodoeshavethecopilottouseittotheiradvantageto

createdashboardsoutofthedata.Aftertimeisup,allthepeopleintheworkshopshouldsendtheirdashboardstoyouforreview.Comparewhichoneswerebetter.Thisissubjective,soconsiderrecruitingotherstovote.

SendasurveyouttothegroupwhohadaccesstothegenerativeBIdashboard.Askthemquestionssuchas:

•Didyouusethenewcopilottobuildyourdashboard?

•Didyoufindithelpful?Ifso,howdidyouuseit?

•Didyougiveupusingitatanypoint?Ifso,why?

•Ifyoudidnotfindithelpful,whatdidyounotlikeaboutit?

•Ifyoudidnotuseit,whynot?

•Didyoutrusttheresultsthatitgaveyou?

•Wereanyresultsinaccurate?Ifso,whatwerethey?

•Wouldyouusethecopilotagaininthefuturetobuilddashboards?

•Onascaleof1to10,howdoyouratethecopilotoverall?

Performinganexerciselikethiscouldhelpidentifytheusefulnessofthecopilotandgiveyourselfsomeobjectivedatathathelpsyoudeterminewhetheryoushouldmoveforwardwithitsadoption.

Trust

Copi

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论