》orkingSmarter:

HowManufacturersAre

UsingArti?cialIntelligence

KeyFindingsandInsightsfrom

Manufacturers

MAY2024

2

WorkingSmarter:

HowManufacturersAreUsingArti?cialIntelligence

May2024

KeyFindingsandInsightsfromManufacturers

NationalAssociationofManufacturers

3

>ALetterfromtheChair

Arti?cialintelligencehasdominatedtheheadlinesinrecentyears,andpeoplearetrulybeginningtograspthepossibilitiesandpowerofthistechnology.ThelaunchofChatGPTandothergenerativeAItoolshasmadethetechnologyevenmore

accessible,puttingitinthehandsofeverydayAmericans.

Manufacturershavebeenattheforefrontofdevelopingandimplementingintelligent

systemsandAItechnologies,includingmachinelearning,deeplearning,natural

languageprocessing,machinevision,digitaltwinsandrobotics.Thishaspositioned

manufacturersuniquelyasbothdevelopersanddeployersofAIinnovations,providinginvaluableinsightsintotheeffectiveandresponsibleuseofthesetechnologies.

AIcanbeaforcemultiplier—andaforceforgood.AtJohnson&Johnson,forexample,AIhasbeenused

effectivelyinanumberofareas,rangingfromthedrugdevelopmentprocesstorestockinghospitals.Ithelpsussortthroughmassiveamountsofdata,yieldinginsightsfortheimprovedhealthandwellnessofpeoplearound

theworld.Itaidsusincreatingtargetedtreatmentsandgettingthemtotherightpatientsattherighttime.Whenweconductclinicaltrials,AIhelpsusmoreef?cientlyestablishsafetyandeffectivenessguardrails,whileallowingustoconducttrialsatalargerscale.AIalsogivesusafarstrongermasteryoveroursupplychains.Overall,it

helpsourpeopledoabetterjoboflivinguptoourcommitmentofimprovinghealthcareoutcomesandmakingourtowns,countryandworldabetterplace.

AIfunctionsbestwithhumansasthecoredecision-makerswithinAI-enhancedprocesses.Theseoperators

mustbeknowledgeable,well-trainedandabletoutilizethetechnologysafelyandtoitsfullestpotential.Early

on,Johnson&JohnsondevelopedanethicalAIframework,aswellasadatascienceacademytoenhanceourteams’digitalacumenandequipthemwithAIengagementskills.AsmanufacturersupskillandtrainmoreteammemberstoworkwithAI,thetechnologywillempowerthoseworkerstobemoreinnovativeandproductive.

Asyou’llreadinthispaper,manufacturersofallsizeshavefoundsimilarwaystouseAItoamplifytheir

operationsandliveuptotheirowncommitments.WithAIsupportingus,manufacturerscandosomuchmoretoimprovethequalityoflifeforeveryone.

Giventheimportanceofthisgenerationaltechnology,policymakersmustdevelopsensible,carefullythought-outframeworksforvariousAIapplications—andtheyshouldleanonmanufacturers’yearsofexperiencetoengineerthoseframeworks.WeneedapolicyenvironmentthatsupportsinnovationandgrowthinmanufacturingAI,

becauseitwillbolsterU.S.competitivenessandleadershipinthiscriticalemerging?eld.

AllpossiblefuturesformodernmanufacturingintheU.S.involveAI.Thistechnologyisagame-changer,anditwillcontinueprovingitselftobeanessentialpartnerontheshop?oor.ThispaperofferspolicymakersawindowintothefutureofAIinmodernmanufacturing—andaroadmaptohelpusgetthere.

KathrynWengel

ExecutiveVicePresidentandChiefTechnicalOperations&RiskOf?cer,Johnson&JohnsonChairoftheBoard,NationalAssociationofManufacturers

4

>KeyInsights

nArti?cialintelligencetoolsareusedwidelyacrosstheindustryandarekeytoadvancingmodernmanufacturing.

nManufacturersareconsumers,developersanddeployersofAIthroughouttheirproductionprocesses.

nThepotentialapplicationsforAIinmanufacturingareexpansiveandcanhelpindustryleadersimproveef?ciency,productdevelopment,safety,predictivemaintenanceandsupplychainlogistics.

nAIreferstoalargeumbrellaoftechnologiesthatincludemachinelearning,machinevisionanddeeplearning.Thesetoolsallowmanufacturerstomaketheirshop?oorssafer,improveworkexperienceandcreate

innovativeproductsthatsolveglobalchallenges.

nManufacturersareimplementingandtestingAIprogramsinawaythatkeepsworkersasthecentraldriversanddecision-makersforAIprocessesorproducts.

nToremainagloballeaderinadvancingAIandsupportingmanufacturinginnovation,theU.S.shouldtakea

cautiousapproachtoAIregulation,tailoranyregulationtospeci?cusecasesandrisks,right-sizecompliance

burdens,supportR&Dandnewworkforcepathwaysandensurethatregulatoryframeworksarealignedglobally.

5

>WhatIsAI?

“WeseeAIasakeystrategicenablerfor

oureffectiveness,to

dothingsbetter,fasterandmoreeconomically,whiledelivering

essentialproductstoourcustomers.”

–SreedharSistu,

VicePresident,AIOffers,SchneiderElectric

Innovationiswhatdrivesmanufacturing,andasaresult,

manufacturershavealwaysbeenattheforefrontofnewtechnologies,strivingtooperatemoreef?cientlyandeffectively.Nowmanufacturersareleadingintheadoptionanduseofarti?cialintelligence.AIisa

broadumbrellaterm,de?nedbytheNationalInstituteofStandards

andTechnologyasa“systemthatcan,foragivensetofobjectives,

generateoutputssuchaspredictions,recommendationsordecisionsin?uencingrealorvirtualenvironments.”1Thesesystemsusedataandhuman-builtalgorithmstosimulatehowhumansperceive,learnand

respondtoquestionsandprompts.AIsystemsareoftenconnectedtoothermachinesandrespondtothedigitalandphysicalworldtosupportprocessesthatcaneitherbeverysimpleorcomplex.2

Whilerecentadvancementsinlargelanguagemodelsandchatbots,suchasChatGPTandGoogleGemini,haveplacedaspotlighton

generativeAItechnology,theseapplicationsrepresentbutafraction

ofthetypesofAIcurrentlyinuse.Infact,manufacturershavebeendevelopinganddeployingintelligent

systemsandAItechnologyformanyyears,intheformofmachinelearninganddeeplearning,natural

languageprocessing,machinevision,digitaltwinsandrobotics,allfurtherexplainedinthefollowingpages.

Theseinnovationsareoftencategorizedunderthebannerof“advancedmanufacturing”or“Manufacturing

4.0.”AIintegrationintomanufacturingprocesseshascontributedalreadytosigni?cantlyimprovedoperationsandtothedevelopmentofnewproducts.3

AIrepresentsatremendousopportunityforthemanufacturingindustry.AItechnologiescanhelp

manufacturersimprovetheiroperationsbyupgradinghowtheyanalyzelargedatasets,identifyingknowledgegaps,providingsolutionsandenablingteamstodevelopnewef?ciencesatscale.WidespreadimplementationofAIacrosstheindustrycouldleadtomoreef?cientprocesses,increasedsustainability,moreinnovative

productsandsaferworkplaces.TheseinnovationswillbothgrowtheeconomyandbolsterU.S.global

leadershipinmanufacturing.GiventhevastpotentialofAI,policyapproachestoAIshouldfurtherthe

developmentofthesetechnologiesandsupporttheirresponsibleusebymanufacturersacrossawiderangeofapplications—strengtheninginnovationand,inturn,supportingU.S.competitivenessontheworldstage.

ThisreportreviewshowAIhasevolvedwithinmanufacturingandhowmanufacturersaredevelopingand

deployingAItechnologiestoinnovatewithintheirbusinessoperationsandacrosstheindustry.ManufacturersarealeadingvoiceontheopportunitiespresentedbyAIandhavemuchtoshareabouttheirexperiences.

ThisreportconcludeswithpolicyrecommendationsthatwouldbestequipthemanufacturingindustrytotakeadvantageoftheimmenseopportunitiesAItechnologyhastooffer.TheNAMsupportsapolicyenvironment

forAIthatencouragessafe,responsibledevelopmentwhilepromotingtheinnovativegrowthofthetechnology.

1NationalInstituteofStandardsandTechnology,Arti?cialIntelligenceRiskManagementFramework(AIRMF1.0)(Washington,D.C.:

DepartmentofCommerce,2023),1,

/nistpubs/ai/NIST.AI.100-1.pdf

.

2TomCulver,LeeGreenandJimRedden,“PeeringintotheFutureofIntelligentSystems,”Research-TechnologyManagement62,no.3(May2019):21-30,

/doi/abs/10.1080/08956308.2019.1587322

.

3HailiZhang,XiaotangZhangandMichaelSong,“DeployingAIforNewProductDevelopmentSuccess,”Research-TechnologyManagement64,no.5(August2021):50-57,

/doi/full/10.1080/08956308.2021.1942646

.

6

>HowDidWeGetHere?

AIinmanufacturingismadepossiblebytheconnectednatureofmachinesandtoolsinmanufacturing

operations.Widespreaddigitalintegration,usingconnectedsensorsandinstrumentstocollectdataacrossshop?oors,enablesmachinelearning,atypeofAIthatwasdevelopedasearlyasthe1980s.4Amachinelearning

systemanalyzesdataandrecognizespatternstotrainitselftomakedecisionsandperformtasksef?ciently.

Deeplearning,anextensionofmachinelearningthatevolvedthroughthe2010s,incorporatesmultiplelayersofreasoninganddataanalysistomimichowthehumanbrainworks.5MachinelearninganddeeplearningarethebasisformostoftheAItoolsmanufacturersuse.AsofOctober2023,74%ofsurveyedmanufacturershad

investedorwereplanningtoinvestinmachinelearning.6

Arti?cialintelligence:atechnologythatcan,foragivensetofobjectives,generateoutputssuchaspredictions,recommendationsor

decisionsthatemulatehumanbehavior

Machinelearning:amodelthatuses

advancedalgorithmstoanalyzedataandrecognizepatternstotrainitselftomakedecisionsandperformtasks

Deeplearning:asystemthatincorporates

multiplelayersofreasoninganddataanalysistomimichowthehumanbrainworks

GenerativeAI:atoolthatusesdeep

learningtocreatecontent,suchastext,imagesorcode,basedondetected

patternsinlargedatasets

4JimDavis,“PuttingIntelligenceBackintoAI,”ManufacturingLeadershipCouncil(Dec.8,2020),

/putting-intelligence-back-into-ai-17349/?stream=all-news-insights

;

MichaelPlatzandShantonWilcox,“AchievingImpactfromEnd-to-EndDigitalization,”ManufacturingLeadershipJournal(January2023),

/achieving-impact-from-end-to-end-digitization-31586/?stream=ml-journal

.

5“Whatisdeeplearning?,”IBM,accessedJan.30,2023,

/topics/deep-learning

.

6PenelopeBrown,“SURVEY:ManufacturersGoAll-InonAI,ManufacturingLeadershipCouncil(Oct.1,2023),

/survey-manufacturers-go-all-in-on-ai-35350/?stream=ml-journal

.

7

>HowDoManufacturersUseAI?

Manufacturersarecollectorsofknowledge.Theybringtogethertheskillsandideasofpeople,sometimesfromallaroundtheworld,tocreatenewproducts.Thesegoodscanbeassmallandsimpleasaboltoraslarge

andcomplexasanautomobile.Themodernshop?oorisinterconnectedandtechnologicallyadvanced.This

enablesmanufacturerstocollectdataabouttheiroperationsandenhancetheirproductionprocesses.Inshort,thetechnologicalinnovationsofmodernmanufacturingenablemachinestoamplifytheproductivepowerof

manufacturingworkers—andAIisthenextstepinthisinnovativejourney.

In2023,theNAM’sManufacturingLeadershipCouncil,aglobalnetworkofexecutivesinthemanufacturing

industry,conductedsurveysonhowmanufacturersuseAIintheiroperationstoexploresomeofthewaysAIisalreadymakinganimpact.WhenaskedaboutwhytheywereinvestinginM4.0technologiesordigitally

integratedinnovationssuchasAI,respondentspointedtocostreduction,operationalawarenessandprocessoptimization,asshowninFigure1.7Thisincludesvisibilityintooperationsbycollectingandanalyzingdatatodevelopinsightintotheperformanceofamanufacturingprocess,andusingdigitaltechnologyanddatato

determineaprocess’sef?ciency,speed,equipmentutilization,materialsusage,waste,etc.,andmakingdecisionsonhowanyofthosefacetscouldbeimproved.

Figure1:WhatAretheMostImportantReasonsYourCompanyInvestsinTransformativeM4.0Technologies?(CheckTopThreeReasons)

72%

51%

41%

32%

22%

21%

19%

14%11%

11%

5%

0%10%20%30%40%50%60%70%80%

Reducecostsandimproveoperationalef?ciencyImproveoperationalvisibilityandresponsiveness ImproveprocessoptimizationandcontrolCompensateforlaborshortages

Improvequality

CreatesustainedcompetitiveadvantageImproveassetreliability

Improvespeedtomarket

ImprovecustomerexperienceMitigateimpactfromdemandswingsDon’tknow

Wherepossible,manufacturersdeveloptheirownAItoolstoaccomplishthesetasks.OtherspurchaseAI

productsfromtechnologycompanies.RegardlessofwhethermanufacturersdevelopAItechnologyordeployAItechnology—orboth—they?ndthatAIhelpseasetheburdenofrepetitivetasks,allowingmanufacturingworkerstodevotetheirenergyandtimetomorecomplicatedandforward-thinkingactivitiesandprojects.

7Ibid.

8

AsdisplayedinFigure2,respondentsnotedawiderangeofusesforAIintheiroperations,thetopthreeofwhichweremanufacturingandproduction,inventorymanagementandqualityoperations/R&D.8

ToimplementAI,manufacturersworktoidentifywhichAIsystemisbestsuitedtohelpthemtackletheir

challenges;howtoresponsiblyandtransparentlycollectthedatanecessarytotrainandruntheAImodel;andwheretoimplementAItotransformcurrentprocesses.Inthiscontext,modernmanufacturersviewdataasa

criticalinputthatcanbeleveragedandutilizedtodiscovernewef?ciencies.AIishelpingtotransformthatdataanddeploysolutionsatascalenotpossibleforhumansalone.

Figure2:WhichoftheFollowingCorporateFunctionsHaveBeguntheAdoptionofAI?(SelectAllThatApply)

ManufacturingandproductionInventorymanagementQualityoperations

R&DIT/OT

Equipmentmaintenance/installation Supplychain ProductdesignDistribution/logisticsSalesandmarketing

FinannceeHumanresourceess

CustomerserviceandsupportLegal

ProcurementSustainability

39%

33%

24%

24%

21%

17%

11%

11%

9%

7%

7%

6%

4%

3%

3%

3%

0%10%20%30%40%50%

Withthelargeamountsofdatacollectedontheshop?oorandthroughouttheiroperations,manufacturersuseAItodesignproductionprocesses,predictivemaintenanceprogramsandlogisticsdecision-makingmodels,

amongmanyotherexamples.ThesecompaniesarepushingtheboundariesofwhatAIsystemscando.ThisputsmanufacturersinauniquepositiontoguidethedevelopmentoftheAIpolicylandscape.

8DavidBrousell,JeffPumaandPaulTate,TheFutureofIndustrialAIinManufacturing(Washington,D.C.:ManufacturingLeadershipCouncil,

2023),

/wp-content/uploads/2023/06/The-Future-Of-AI-In-Manufacturing-MLC-2023.pdf

.

9

Ef?ciency

Thecontinued,expandedimplementationofearlyAItechnologies,suchasmachinelearning,hasimproved

manufacturers’ef?ciency.Greateref?ciencyallowsmanufacturerstoallocateresourcesinamorecost-

effectivemanner,improveshop?oorprocesses,implementmoresustainablepracticesanddiscovernew

opportunitiesforgrowth.IninterviewsconductedbytheNAM,onechemicalproductioncompanystatedthattheapplicationofmachinelearningtodatacollectedfromchemicalreactorsenablesoperatorstomakebetterdecisionsabouthowtooperatethem.TheAImodelalertsoperatorswhenitisoptimaltomakechangesintheprocess,ratherthanmanuallykeepingtrackofallsensorsordependingonaspeci?coperationaltimetable.

Humanoperatorsarestillatthecenterofdecision-makingandoperations,butAIhashelpedimprovethereliabilityoftheirprocessesandthequality,deliveryandsafetyoftheirproducts.

“HitachiisfocusedonapplyingAI,machinelearningandrelatedtechnologiestowardaddressingreal-worldchallengesinindustrialandsocietaldomains.

Functionalareasincludemaintenanceandrepair,operationsoptimization,qualityassurance,safetymanagement,supplychainmanagementandautomationand

control,amongothers.Thegoalistheend-to-endoptimizationofkeyindustrialprocesses.”

–ChetanGupta,GMoftheAdvancedAIInnovationCenter,Hitachi,Ltd.andHeadoftheIndustrialAILabatHitachiAmerica

MachinevisionisamorerecentadvancementinAI,enablingindustrialequipmentto“see”bygatheringand

analyzingvisualdatainitsenvironmenttoformconclusions.Almost80%ofMLCsurveyrespondentshad

investedorplannedtoinvestinvisionsystems.9Onelogisticscompanyusesmachinevisiontosortpackages,dependingonAI-enabledrobotstorespondtoandmakedecisionsbasedonever-changingsituationsand

conditionsratherthansimplyrepeatingthesamepatterns.Manycompaniesarealsousingmachinevisionto

performqualitycontrol,quicklyreviewingpartsandmaterialsfordefectsthataremoredif?cultforhumansto

detect.Forsomemanufacturers,thishasbeenaparadigmshiftandhasallowedtheirmachinesandprocessestoworkmoreef?cientlyandrespondtonewsituationsandproblemsmorequickly.

AImodelscanalsoperformpredictiveanalyticsusingdatacollectedfromdigitallyenableddevices.Aresult

ofthisanalysiscanbepredictivemaintenance,oridentifyingpartsthatarenotperformingef?cientlysothat

theycanbereplacedbeforetheybreak.MorethanhalfofMLCsurveyrespondentsstatedthatpredictive

maintenanceisakeyAIapplicationintheiroperations.10Suchef?cienciescanpreventunplanneddowntimeforproductionaswellasenhancesustainabilitybyenablingmoreenergy-ef?cientprocesses,loweringwasteanddecreasingemissions.

9Brown,“SURVEY:ManufacturersGoAll-InonAI.”10Ibid.

10

Safety

“AIgivesustheabilitytocombinedigitalandphysicalteamsandtohelpourpeoplebyreducingrepetitivetasksandphysicalstresswhile

promotingsafety.”

–JoelStenson,SeniorVicePresidentofOperations

Technology,UPS

ManufacturersareinterestedinusingAItoimprovesafety

foremployeesandoperations.OneautomotivemanufacturerisusingAIandmachinevisiontomonitorintersectionsof

productionlanes,lettingworkersknowifaforkliftorother

machineryiscomingaroundthecorner,outsideofthe

peripheryoftheirvision.ThisuseofAIhelpstoprevent

humanmistakesandgreatlyimprovessafetyontheshop

?oor.Forindividualworkers,ergonomicassistance,likea

roboticexoskeletonthatcollectsandlearnsfromdataon

thewearer’smovementsandtheenvironment,canenhancehumanstrengthandpreventinjuries.Overtime,thesetoolshavethepotentialtoimproveworkersafety,makejobslessphysicallydemandingandreducehealthcarecosts,whicharekeyconcernsforemployers.

OthercompaniesuseconnecteddevicesandAImodelingtoimprovethecustomerexperience.Infact,47%ofsurveyedmanufacturersplantodeploymorecustomer-facingAItoolsinthenexttwoyears.11Oneautomotivecompanyhasalreadybeguncollectingdatafromtheirnewestvehiclemodelsandalertingcustomerswhen

theAIidenti?eschangesintheperformanceofthevehicle,allowinguserstopreventcostly,inconvenientandpotentiallydangerousproblemslater.

ProductDevelopmentandDesign

Researchershavefoundthatproductdevelopmentcanbene?tfromusingAImodelstolearnfromaccumulateddataandthathigherAIusageinthedevelopmentprocessincreasessuccess.12ManufacturersinmanydifferentindustrysubsectorsareutilizingAItodevelopnewproducts.AsupplierofautomotivepartshascreatedanAI

toolthatprocesseswheelgeometrydata,allowingthecompanytomorequicklydevelopwheeldesignsthatperformbetterandareproducedmoreef?ciently.AIallowsthiscompanytobringproductstomarketfaster,

respondtodesignchangesmorequicklyandbetterapplytheknowledgeoftheirengineeringteam,unlockingcontinuousinnovationandlearning.

OnepharmaceuticalcompanysharedthatitisusingAImodelstoidentifynewwaystodevelopmoleculesand

advanceindividualizedtreatmentsfordisease.ThiscompanyisdevelopingtheirownAImodelsto?ndmorepreciseendpointsfortreatments,whichmakeclinicaltrialssafer,moreeffectiveandwithagreatermarginforsuccess.

Training

Manufacturerswanttokeeptheworkofpeople,notcomputers,atthecenteroftheiroperations.TheingenuityofworkersisakeycomponentinanyAIprocessesorproducts.WhenitcomestoAIinmanufacturing—workersarethedriversanddecision-makers,andAIworkstomaketheirjobseasierandmoreef?cient.ManycompanieshavefoundthattheirbestAIoutcomesresultfromenrichingemployees’experienceatwork,thusimproving

theiroutput.AItechnologyisoftenusedtocomplementandaugmenttheworkofhumans,likeaco-pilot.

Thisapproachenhancesworkeref?ciencywhilestillprioritizinghumanexperienceandingenuity—ultimately

11Ibid.

12Zhang,ZhangandSong,“DeployingAIforNewProductDevelopmentSuccess.”

11

increasingtrustandcon?denceinAIsystems.13Themosteffective

AImodelsarehuman-centered,allowingthemtolearnandunlearn,

continuouslyimprovingtomeettheneedsoftheirhumanoperators.14

Manufacturersareupskillingtheirworkforceactivelytomeetthe

opportunitiesofAI.Manycompanies,fromlogisticstopharmaceuticals,aresettinguptrainingprogramstohelpemployeesdeveloptheir

con?denceandcompetencyintheuseofAIsystems.Thesetraining

programsincludeafocusonsafetyandcontrol,tolimitriskstoworkersandtoprotectcompanies’intellectualpropertyinthefaceofthe

increasedcybersecurityrisksthatcomefromaninterconnectedshop?oor.

Recruitmentofnewemployeeshasalsochanged.Oneautomotive

suppliernotedthattheyhaveincreasedtheirhiringofthedatascientistsnecessarytobuildandimplementAIsystems.

ThemosteffectiveAI

modelsarehuman-

centered,allowingthemtolearnandunlearn,

continuouslyimprovingtomeettheneedsof

theirhumanoperators.

Attheendofworkers’careers,companiescanuseAIsystemstomanageknowledgeaskeypersonnelenter

retirement.Asof2019,nearlyone-quarterofthemanufacturingworkforcewasover55,15andoneelectri?cationandemergingtechnologycompanyisusingAI-utilizingsystemsforknowledgemanagementandtotrainnew

employeesontheskillsoflegacyworkers.TheyarealsousingAImodelstoidentifyfuturechallengesand

predictwhichskillswillbeneeded,identifyinghowmanyemployeeswillneedtobetrainedinthesenewskills.

SupplyChain

ManufacturersareusingAImodelstopredict,preventormitigatedisruptionsintheirsupplychainsandmake

moreinformeddecisionsabouttheirlogisticsplans.Thesetoolsallowtheindustrytobemoreresilienttorisk,

preventproductionstoppagesorshortagesandeffectivelydeliverproductstocustomers.ThisisagrowingareaofAIimplementation,with21%ofsurveyedmanufacturersalreadyusingAIintheirsupplychainmanagement

and60%planningtodeployitinthenext12–24months.16

13JimEuchner,“Littleai,BigAI—GoodAI,BadAI,”Research-TechnologyManagement62,no.3(May2019):10-12,

/doi/full/10.1080/08956308.2019.1587280

.

14Davis,“PuttingIntelligenceBackintoAI.”

15TheManufacturingInstituteandAlfredP.SloanFoundation,TheAgingoftheManufacturingWorkforce(Washington,D.C.,July2019),

/research/the-aging-of-the-manufacturing-workforce/

.

16Brown,“SURVEY:ManufacturersGoAll-InonAI.”

12

Inapplication,AItoolsinthesupplychaincanhelpcompaniesmakethemostoftheirinventories.One

technologyandcomputingcompanybuiltandintegratedanAIprogramthatusedmachinelearningtoassessthecommonpartsacrossdifferentareasoftheiroperations,identifyingwhenandwherepartscouldbeshiftedfromoneareatoanotherasneeded.Becausesparepartshadbeenscrappedpreviouslyifunused,using

machinelearningtoassesshundredsofpartsandgreatlyreducedpotentialwaste—anunfeasibletaskforahumanteam—hassavedthiscompanymillionsofdollarsinreplacementandsourcingcosts.ThecompanyhopestoexpandthisprojecttoincludefurtheradvancementsinAItechnology,includingdeeplearningandgraphneuralnetworks.

Intheaerospacesector,AI,speci?callymachinelearning,hasevolvedtodeliversolutionsintheareasofautonomy,suchasobstacleavoidanceandautomaticvehicletaxiing,aswellasoptimizationofcargoinairplanesthatare

connecteddigitally,andmanyotherareas.Theseadvancements,leveragingdeeplearning,continueprovidingsolutionstotechnologygapsthatwouldotherwisebehardtoachievewithtraditionalapproaches.

Morewidely,usingmodelsthattakeglobalandnationaltrendsanddisruptionsintoaccount,manufacturerscanmakeswiftchangestotheirsupplychaindecisions.Thisallowsthemtobemoreresilienttounexpectedchanges,withmodelsabletoidentifynewshippingpatterns,alteredsupplierchoicesorinventoryshocks.

>HowAreManufacturersTestingAISystems?

AsmanufacturersexpandtheuseofAItechnology,theyare?ndingmorerobustwaystoguaranteethatthesetechnologiesaresafeandreliablefortheiremployeesandcustomerstouse.ManycompaniesareapproachingAIthroughthesamekindofprovenrisk-managementframeworksthattheyusefortheirITandcybersecurity

programs.Companiesarealsodevelopingtheirowninternalgovernanceprograms.

Manufacturersareusingtestinggroups,bringingtogetherAI,ITandoperationsprofessionals,toidentifywhere

algorithmsmightbeinaccurateandtovalidatethattheirsystemsmeethighthresholdsofsuccess.WhentestingnewAIsystems,oneshippingandlogisticscompanyfoundthatinternalfacilitysafetyteamsandtheirthird-partytestingorganizationsbothneededtodevelopanewknowledgebaseandupskilltogether.

Inthiswayandothers,manufacturersarebuildingtheirowngovernanceprogramsfordataandAIsystems,

maintainingdataprivacyandconductinginternaltestingbeforenewprogramsaredeployed.Thisistrue

especiallyforheavilyregulatedindustries,suchasautomotive,pharmaceuticalsandaerospace,thatalready

mustmeetmanyofthesafetybenchmarksapplicabletothedevelopmentandtestingofsafeAIsystems.Manyareworkingdirectlywiththegovernmentalreadytodevelopcerti?cationsforcriticaltechnologiesthatdonot

disrupttheirdeploymentofAI.

>WhatShouldWeDoNow?

ManufacturersarecommittedtotheresponsibledevelopmentanddeploymentofAI.AIhasbecomecritical

tomodernmanufacturing,andAItechnologiesandcapabilitiesar