《世界经济论坛(WEF):2022释放人工智能在制造业中的价值白皮书(英文版)(26页).pdf》由会员分享,可在线阅读,更多相关《世界经济论坛(WEF):2022释放人工智能在制造业中的价值白皮书(英文版)(26页).pdf(26页珍藏版)》请在三个皮匠报告上搜索。
1、Unlocking Value from Artificial Intelligence in ManufacturingW H I T E P A P E RD E C E M B E R 2 0 2 2In collaboration with MEXT Technology CenterContentsForewordExecutive summaryIntroduction1 Unlocking value in manufacturing through AI2 Shedding light on common barriers to industrial AI adoption3
2、A collection of AI applications in manufacturing4 A step-by-step approach to implementing scalable industrial AI applicationsConclusionContributorsEndnotes345681117212225Cover:Jian Fan,Getty Images Inside:Getty Images 2022 World Economic Forum.All rights reserved.No part of this publication may be r
3、eproduced or transmitted in any form or by any means,including photocopying and recording,or by any information storage and retrieval system.DisclaimerThis document is published by the World Economic Forum as a contribution to a project,insight area or interaction.The findings,interpretations and co
4、nclusions expressed herein are a result of a collaborative process facilitated and endorsed by the World Economic Forum but whose results do not necessarily represent the views of the World Economic Forum,nor the entirety of its Members,Partners or other stakeholders.Unlocking Value from Artificial
5、Intelligence in Manufacturing2ForewordTrkiye has established itself as a key global player in advanced manufacturing and aims to boost its position through Fourth Industrial Revolution technologies.In recent decades,the country has made significant efforts to position itself as a global innovation h
6、ub,excelling in developing state-of-the-art technologies in ground-breaking companies in various fields.Artificial intelligence(AI)technology applications are part of this effort.In principle,AI could unlock more than$13 trillion in the global economy and boost GDP by 2%per year.1 However,companies
7、struggle to tap into the value that AI applications can create.This paper seeks to uncover the hidden potential of AI in the manufacturing sector and the respective end-to-end systems by providing practical use cases and critical enablers to help harness its potential.Coupled with the energy crisis
8、and material shortages facing the world,manufacturing players need to go beyond traditional operating methods to drive efficiency and sustainability.The twin challenges of technological progress and socio-political distress call for new forms of cooperation that respond to heightened demand for loca
9、lization while recognizing the drivers of connectivity that shape global impact.Acknowledging this,the Centre for the Fourth Industrial Revolution in Trkiye mandated by the Ministry of Industry and Technology and established by the Turkish Employers Association of Metal Industries(MESS)joined the Wo
10、rld Economic Forums Centre for the Fourth Industrial Revolution Network,the foremost platform helping leaders anticipate emerging technologies and drive their inclusive and sustainable adoption.The network links on-the-ground experience and action with global network-based collaboration,learning and
11、 scaling.This white paper is an output of the ongoing partnership between the Forums Platform for Shaping the Future of Advanced Manufacturing and Value Chains and Platform for Shaping the Future of Technology Governance:Artificial Intelligence and Machine Learning,the Centre for the Fourth Industri
12、al Revolution Affiliate in Trkiye and MESS.It highlights case studies from organizations on the impact,feasibility and scalability of AI in manufacturing.It identifies several opportunities and lessons from the community on how to increase operational efficiency,sustainability and workforce engageme
13、nt in manufacturing and value chains by using AI.We hope this report will provide decision-makers with a better understanding of how to unlock the untapped potential of industrial artificial intelligence(AI).We look forward to collaborating with you to deploy these technologies responsibly.Unlocking
14、 Value from Artificial Intelligence in ManufacturingDecember 2022zgr Burak Akkol Chairman,Turkish Employers Association of Metal IndustriesJeremy Jurgens Managing Director,World Economic ForumUnlocking Value from Artificial Intelligence in Manufacturing3Executive summaryRecent global developments an
15、d an ever-growing list of shocks and disruptions have put further strain on already shaken global value chains.The complexity of current challenges impacting manufacturing and value chains calls for the need to go beyond the traditional means of driving productivity to uncover the next wave of value
16、 for businesses,the workforce and the environment.Artificial intelligence(AI)is a crucial enabler of industry transformation,opening new ways to address business problems and unlock innovation while driving operational performance,sustainability and inclusion.Even though the impact of AI application
17、s on manufacturing processes is known,the full opportunity from their deployment is still to be uncovered due to a number of organizational and technical roadblocks.Recognizing this need,the Centre for the Fourth Industrial Revolution Trkiye,together with the World Economic Forums Platform for Shapi
18、ng the Future of Advanced Manufacturing and Value Chains and Platform for Shaping the Future of Technology Governance:Artificial Intelligence and Machine Learning,convened industry,technology and academic experts to shed light on these challenges and propose a step-by-step approach to overcome them.
19、The consultations revealed six main challenges hindering the adoption and scaling of AI applications in manufacturing:1.A mismatch between AI capabilities and operational needs2.The absence of a strategic approach and leadership communication3.Insufficient skills at the intersection of AI and operat
20、ions4.Data availability and the absence of a data governance structure5.A lack of explainable AI models in manufacturing6.Significant customization efforts across manufacturing use casesThe consultations show that leading manufacturers have successfully overcome the challenges mentioned above,implem
21、enting a variety of AI applications and achieving a positive impact on operational performance,sustainability and workforce engagement,mainly in six areas:health and safety,quality,maintenance,production processes,the supply chain,and energy management.While opportunities enabled by AI in manufactur
22、ing are promising and attracting many leaders,organizations are looking for a common framework that outlines how to implement AI solutions and ensure a successful return on investment.Based on the consultations,this white paper presents one step-by-step process as an example of how it is possible to
23、 overcome barriers,using the AI Navigator2 developed by the INC Invention Center as a reference:Phase 0:Initiation to build the fundamentals strategy,data and workforcePhase 1:Ideation to identify potential use cases and conduct a pre-selectionPhase 2:Assessment to select use cases and identify prio
24、rities via gap analysisPhase 3:Feasibility to complete all required tests and studiesPhase 4:Implementation,which requires iteration and piloting using agile project managementMoving forward,the World Economic Forum and the Centre for the Fourth Industrial Revolution Trkiye will continue to work clo
25、sely with stakeholders in the Centre for the Fourth Industrial Revolution Network and across industries to accelerate the journey to capture value from AI in manufacturing globally.It will offer the Turkish Employers Association of Metal Industries(MESS)Technology Centre as a unique testing and coll
26、aboration system for businesses to pilot new AI applications and foster a collaborative approach among a diverse group of stakeholders to ensure the right AI capabilities are built in manufacturing and rolled out worldwide.Unlocking Value from Artificial Intelligence in Manufacturing4IntroductionCom
27、panies across value chains are now facing an energy crisis and material and key component shortages,even as they are still recovering from and adapting to COVID-19 impacts.The complexity of the challenges impacting operations calls for the need to go beyond the traditional means of driving productiv
28、ity to uncover the next wave of value and address sustainability and workforce challenges.Artificial intelligence(AI)can enable a new era in the digital transformation journey,offering tremendous potential to transform industries to gain greater efficiency,sustainability and workforce engagement by
29、generating new insights from large amounts of data.However,despite this promising value creation potential,the deployment of AI in manufacturing and value chains is still below expected levels.Based on a global survey conducted over the last four years of more than 3,000 companies across industries
30、and geographies,a growing number of companies recognize the business imperative to improve their AI competencies:70%of respondents understand how AI can generate business value 59%have an AI strategy in place 57%affirm that their companies are piloting or deploying AI.Despite these trends,only 1 in
31、10 companies believe they generate significant financial benefits with AI.3 While manufacturers acknowledge the importance and urgency of embedding AI in their processes and while leading companies have already internalized it in their business processes,many are becoming disillusioned with their ef
32、forts to capture value from it and lag in developing the right AI capabilities.Understanding the purpose and role of AI is key to solving manufacturing challenges.With a problem-oriented approach,AI efforts can be linked to clear business targets,giving business units and business functions a joint
33、interest in making the transformation successful.4This white paper sheds light on the benefits that can be achieved through industrial AI and the successful AI applications implemented across industries,lessons learned and tangible impacts.Consultations conducted with the multistakeholder initiative
34、 community find that industrial AI helps people work in a smarter,safer and more efficient way.However,to unlock its full potential,companies require an understanding of current barriers to adoption and a structured approach to overcome them.Therefore,this paper also presents one example of a step-b
35、y-step guide to successfully implementing scalable industrial AI use cases.Unlocking Value from Artificial Intelligence in Manufacturing5Unlocking value in manufacturing through AI1The artificial intelligence(AI)revolution allows the conversion of large amounts of data into actionable insights and p
36、redictions that can provide impetus to data-driven processes.Manufacturing companies capture value from AI using different mechanisms,the most common being eliminating redundant work,solving existing problems and revealing hidden value by analysing and recognizing patterns in data.AI is applied to a
37、ugment tasks such as classification,continuous estimation,clustering,optimization,anomaly detection,rankings,recommendations and data generation to solve industrial problems.5 Consultations with senior executives from the World Economic Forums Platform for Shaping the Future of Advanced Manufacturin
38、g and Value Chains and Platform for Shaping the Future of Technology Governance:Artificial Intelligence and Machine Learning,as well as members and partners of the Centre for the Fourth Industrial Revolution Trkiye,find that AI can help drive a step-change in manufacturing,yielding significant benef
39、its in three categories(figure 1):Operational performance by automating and optimizing routine processes and tasks,increasing productivity and operational efficiencies,improving quality(e.g.reducing defects,forecasting unwanted failures)and optimizing production parameters Sustainability by optimizi
40、ng material and energy usage,increasing energy efficiencies,reducing scrap rates and extending machine lifespans Workforce augmentation by guiding the decision-making process and parameter setting,enhancing the accuracy of predictions and forecasting,reducing repetitive tasks and increasing human-ro
41、bot interactionsAI applications in manufacturing help increase operational performance,drive the sustainability agenda and empower the workforce.Unlocking Value from Artificial Intelligence in Manufacturing6Performance(e.g.yield optimization)Throughput(e.g.fewer unwanted breakdowns,decreased lead ti
42、me)Quality(e.g.fewer process defects and failure rates)Business uptime(e.g.productive time and capacity)Operational performanceDecision-making and planning supportCollaborationPrediction and forecasting accuracyTask automationWorkforce augmentationRisk(e.g.feedback mechanism to avoid incidents and a
43、larms)Material efficiencyEnergy efficiency(e.g.energy savings and thermal efficiency)Machine lifetimeScrap rate and used materialSustainabilityDimensions of value creation with AI in manufacturingFIGURE 1Unlocking Value from Artificial Intelligence in Manufacturing7Shedding light on common barriers
44、to industrial AI adoption2Implementing AI solutions requires continuous project management efforts,expectation management andthe necessary resources.Despite this potential,companies have not yet fully realized the vision of AI-powered manufacturing systems.To unlock the untapped value of industrial
45、AI,pinpointing the source of a companys struggles and defining the roadblocks open a new path to think through and derive the right solutions to overcome them.As the barriers to AI adoption stem mainly from organizational,strategic and technical components,understanding them will help identify a pat
46、hway to implement scalable AI applications.Consultations with the community of over 35 senior operations executives,technology experts and academics have identified six challenges hindering the adoption of AI in manufacturing and value chains(figure 2).Barriers to AI adoption in manufacturingFIGURE
47、2Mismatch between AI capabilities and operational needsAbsence of a strategic approach and leadership communicationInsufficient skills at the intersection of AI and operationsData availability and absence of a data governance structureLack of explainable AI models in manufacturingSignificant customi
48、zation efforts across manufacturing use casesUnlocking Value from Artificial Intelligence in Manufacturing8Manufacturers have often selected AI projects based on existing technical capabilities instead of focusing on the impact on business operations.The match between business pain points and AI tec
49、hnologies is not always thoroughly considered.Therefore,AI solutions may be technically feasible but fail to solve a relevant,impactful problem in operations.This causes a mismatch of expectations and hinders their wider adoption in manufacturing.Building a solid business case with a problem-oriente
50、d approach that clearly defines business needs and evaluating the value of an AI solution compared to alternative solutions are the first steps in overcoming that barrier to adoption and scale.Mismatch between AI capabilities and operational needsA clear company-wide AI strategy and communication pl
51、an are often ignored.Without the right sponsors and committed leaders to start the dialogue and collect the buy-in from end-users,the onboarding of AI applications across the company cant occur due to workforce reluctance.As AI is changing the ways of working,communicating the strategic approach,ben
52、efits and new processes can help increase end-users willingness to embrace it in their routines.Absence of a strategic approach and leadership communicationExternal consultants or information technology(IT)experts who have a limited understanding of the manufacturing requirements on the shop floor o
53、ften lead AI projects.However,to be successful,AI applications require development and implementation by cross-functional teams with diverse expertise at the convergence of IT,operational technology(OT),data and AI technologies.This requires upskilling the workforce and attracting new talent in manu
54、facturing.Insufficient skills at the intersection of AI and operationsUnlocking Value from Artificial Intelligence in Manufacturing9Applying machine learning models requires training on large amounts of data to recognize patterns and relationships.6 However,manufacturing companies often rely on smal
55、l data sets and fragmented data,hindering the accuracy of the resulting insights.Even when available,these data sets may not represent appropriate failure cases or relevant process situations and are mostly not interoperable.Creating a single source of information ensures that businesses operate bas
56、ed on standardized,relevant data across the organization.To overcome this challenge,sharing data across companies boundaries can support joint efforts to adopt artificial intelligence techniques in the manufacturing sector and rely,in turn,on a set of organizational and technological success factors
57、.7Data availability and the absence of a data governance structureThe perception of AI models as complex,non-transparent and uninterpretable systems hinders their deployment.Manufacturers need AI models that are either open and transparent to build trust in the predictions and specific results or in
58、terpretable for domain experts to accept them.AI-provided predictions need to be meaningful,explainable and accurate and have a warning mechanism in place to minimize risks.Explainable AI tools and techniques allow experts to obtain justifications for their results in a format that manufacturing use
59、rs can understand.The greater the confidence in the AI-powered output,the faster and more widely AI deployment can happen.Lack of explainable AI models in manufacturingFactories are complex engineered systems and AI models need configuration to be adapted to each process and conform to its constrain
60、ts.Hence,it is not possible to simply apply trained AI models or pipelines from one manufacturing use case to another.The design of the machine learning pipeline and the pre-processing,training and testing of AI models still need manual intervention for customization,which is not yet fully automated
61、.Additionally,industrial companies struggle to find commercially available hardware and software with off-the-shelf AI features that require minor customization.Significant customization efforts across manufacturing use casesShedding light on these challenges and understanding them can help identify
62、 the right solutions and approaches to overcome them.Unlocking Value from Artificial Intelligence in Manufacturing10A collection of AI applications in manufacturing3AI applications can boost operational performance and lead to a positive impact on sustainability and workforce engagement.Consultation
63、s with over 35 senior operations executives and technology experts find that leading manufacturing companies have successfully managed to approach and overcome the challenges mentioned above by starting with their business needs,outlining a clear strategy,building cross-functional capabilities and p
64、utting a stronger focus on data governance,and selecting AI models that meet their needs.They have implemented a variety of AI applications that have boosted their operational performance and led to a positive impact on sustainability and workforce engagement.To illustrate the potential and feasibil
65、ity of AI in manufacturing,the creation of an industrial AI use case library with input from the community has started.The 23 use cases collected across different industries cover six main application areas:health and safety,quality,maintenance,production process,supply chains,and energy management(
66、figure 3).Unlocking Value from Artificial Intelligence in Manufacturing11AI in manufacturinguse casesHealth and safetyEmployee health&safety:incident prevention Process safety:advanced alarm analyticsQualityQuality inspection in assembly Quality assurance/defect inspectionQuality testingQuality pred
67、iction MaintenanceMachine health monitoring:predictive maintenanceMaintenance planning Production processProcess optimizationLine balancingProduct design and developmentProcess parameter optimizationProduction planning/decision supportSupply chainsFuture demand and price forecastingSupply chain cont
68、rol towerWarranty and service managementEnergy managementEnergy optimizationElectricity demand forecasting Heating and cooling optimization123456Leading manufacturers are implementing a variety of AI applicationsFIGURE 3Source:Company interviewsThe use cases collected provide valuable insights indic
69、ating the business need,the solution implemented and the impact achieved.The applications show that the return on investment(ROI)is positive and the payback period of the investments is usually tangible within 1-2 years.After piloting the AI applications in one division,manufacturing companies eithe
70、r have already deployed to multiple divisions or have the vision to scale.Unlocking Value from Artificial Intelligence in Manufacturing12A collection of AI in manufacturing use casesTABLE 1Use caseCompanySectorAI applicationImpactHealth&safetyProcess safety:advanced alarm analyticsTpra,TrkiyeEnergyM
71、odel designed as an experienced operator/engineer in continuous estimation and classification of alarms,detection of nuisance alarms,alarm flood analysis and recommendation of better configurations.Root causes,next-best actions and set points extracted from the historical data through basic descript
72、ive analytics and data science pre-process techniques Total time of alarm floods decreased by 40%Number of alarms decreased by 50%Time efficiency:Alarm rationalization meetings shortened from 4 hours to 30 minutesEmployee health&safety:incident preventionIntenseye,USAManufacturingImage recognition b
73、y monitoring the shop floor with existing cameras,receiving real-time alert notifications and enhancing employee health and safety(EHS)to eliminate life-altering injuries Unsafe situations and actions reduced by 70-80%With a safer environment,a more productive workforce with increased business uptim
74、e createdQualityReal-time spot weld quality prediction Martur Fompak,TrkiyeAutomotiveExamining the effective parameters on the frames being welded in robotic spot weld stations(weld quality)and predicting the spot nugget diameter realized in line in real time Up to 40%savings achieved in energy use
75、Scrap rate reduced while ensuring sustainability in production Costs reduced by 60%by preventing the use of excess welding materialsDetection of carbon coating defectsBosch,TrkiyeAutomotiveVisual inspection to ensure the coating quality is good by checking parts and searching for coating defects in
76、four different classes:scratches,damages,black in black,silver Productivity increased by 11%15 million parts checked had no incidentsQuality assurance with federated learning in controlHuawei,ChinaProductionOptimizing quality inspection of customized products by deploying cloud services and a federa
77、ted learning approach(local data collected,global optimum interpolated and in turn shared back to all local facilities without disclosing sensible product or process data)Productivity increased by 30-40%Lead time reducedQuality inspection in assembly verificationEthon AI,SwitzerlandElectronicsExplai
78、nable computer vision methods used to support factory workers in detecting assembly errors on printed circuit boards(e.g.missing,faulty,or wrong components)via a human-AI interface(camera system with live feedback)10 x less implementation effort expended Trustworthiness of the system increased with
79、the explainable modelUnlocking Value from Artificial Intelligence in Manufacturing13Use caseCompanySectorAI applicationImpactQualityQuality testingKarsu,TrkiyeTextileVisual inspection of fibre ratio in yarn content using microscopic images to check production quality and to analyse customer complain
80、ts Report preparation time for customer complaints and analysis expected to decrease by 90%Expert requirement for the subject will be eliminated Quality inspection in drug-and patient safetyKrber Digital,GermanyPharmaceuticalsVisual inspection of the quality of pharmaceuticals while AI recognizes pa
81、tterns instead of measuring physical image values,which decreases the false-reject of products Reduction of false-reject rate by an average of 88%Detection rate increased by an average of 38%Approximately 2x faster time-to-market achieved(transferability)in vision setupPredictive qualitySchneider El
82、ectric,FranceElectronicsAn AI engine that predicts the demagnetization voltage to reduce the number of iterations during relay tests in residual current device product range Machine capacity increased Capex investment reduced Rejections reducedQuality predictionObeikan Digital Solutions,Saudi Arabia
83、ChemicalsThrough combination of digital twin and innovative AI,process anomaly conditions and drivers detectedStatistical process control algorithm,a proven approach of quality control,used Productivity and quality sustainability increased Overall equipment effectiveness in PET lines improved by 20%
84、Customer complaints reducedProduction processProcess optimizationFero Labs,USASteel Providing automated software to take preventive actions early in the production process with explainable AI models to reduce raw material use and minimize costs and emissions during steel production Alloy use reduced
85、 by 9%at steel mills Failure rate eliminatedLine balancingKhenda,TrkiyeAutomotiveAI-based video analytics to label the actions of manual tasks to eliminate operator-related errors and improve manual manufacturing processes and optimize line balancing Productivity increased by 25%By increasing qualit
86、y and efficiency,error costs eliminated and waste and defective products avoidedProduction parameter optimizationDataprophet,South AfricaFoundryGenerating insights into the complex interactions between hundreds of process parameters and their impact on final quality by using deep learning algorithms
87、Application then prescribes next-best step to optimize production without poor quality Defects reduced to 0%from a 6%of historical defect rate Number of quality stops reduced from 81 to 20 per weekUnlocking Value from Artificial Intelligence in Manufacturing14Use caseCompanySectorAI applicationImpac
88、tProduction processAdvanced decision support system on performance testArelik,TrkiyeHome ApplianceImproving cooling test performance in different and dynamically changing climatic conditions to shorten the test duration by an in-house decision-making system based on AI and machine learning(ML)Servic
89、e call rate improved by 15.3%.17.8%of test capacity increased by decreasing the test time from 80 min.to 65 min.17.8%in energy savings per unit in LPT(long performance test)system 16.7%of warranty cost improvement per unit achievedProcess managementGEP,USAChemicalsImplementing AI-enabled process con
90、trols to manage catalyst ingestion based on pressure and temperature changes in the reactor and to manage the transfer rates Overall batch cycle time reduced by 22%and need to add new reactor capacity alleviatedMaintenancePredictive maintenance:machine health monitoring and maintenance planningSense
91、more,TrkiyeCement Detecting the machinery failure mode by collecting the continuous vibration data through fault estimation,early warning and maintenance planning with AI on fans and electric motors Downtime reduced by 90%Maintenance cost reduced by 25%Machine life increased by 20%Operation producti
92、vity improved by 25%Predictive maintenanceThe Center for Intelligent Maintenance Systems(IMS),USAElectronicsPredicting the concentration of contaminative particles before it can negatively impact the production yield,allowing chamber cleaning to be performed in a proactive mannerSolution is for dry
93、etching chamber in semiconductor manufacturing to monitor the deposit accumulation process inside the processing chamber 70%reduction in the costs of unplanned downtime Competitiveness improvedPredictive maintenancePredictronics,USAAutomotiveAI-based predictive solutions for industrial robots for an
94、 automotive manufacturing client to monitor welding robot health and ultimately predict and prevent failure events and schedule maintenance,saving time,money and resources 50%reduction in unplanned downtime Inefficient maintenance practices reducedUnlocking Value from Artificial Intelligence in Manu
95、facturing15Use caseCompanySectorAI applicationImpactSupply chainFuture demand and price forecastingSmartOpt,TrkiyeChemicalForecasting future demand for products,services and raw material prices by automatically training AI model and tuning the model parameters automatically without user input Foreca
96、sting accuracies for the next 6 months reached 85-99%Warranty and service managementTofa,TrkiyeAutomotiveComponent-based prediction granularity as the components of the vehicles affected by different factors to dwetermine warranty expenditure of the coming years for the sold vehicles Prediction accu
97、racy increased from 70%(manual prediction)to 95%,which resulted in reduction of reserve fund by 10%per yearEnergy managementEnergy optimizationCanvass AI,CanadaFood and BeverageAnalysing thermal efficiencies,ambient conditions to control real-time natural gas optimization and consumption targets acr
98、oss turbines;electricity production targets and steam demand from downstream boilers 5.09%gain in thermal efficiency,which translates to 9M lbs/yr and energy cost savings achieved CO2 emissions reduced Overall equipment effectiveness improvedElectricity demand forecastingFord Otosan,TrkiyeAutomotive
99、Predicting electricity usage with changing periods in the 12 different regions of the factory with a model that worked automatically in predetermined periods and eliminated error factors Prediction accuracy score varies between 80-95%for the total consumption of the factory The pre-emptive right to
100、buy the electricity is reserved with the electricity providerHeating and cooling optimizationMakinarocks,Republic of KoreaElectric vehicleOptimizing energy management system(vehicle temperature)of electric vehicles by simulating the control environment using a deep neural network-based dynamics mode
101、l and implementing a reinforcement learning method to improve energy efficiency through optimized control inputs Improved the energy management systems energy efficiency in electric vehicles by 10%on average,with a maximum increase of 25%An additional 5-7%increase in driving distance per every 10-15
102、%increase in energy management system energy efficiencySource:Company interviewsUnlocking Value from Artificial Intelligence in Manufacturing16A step-by-step approach to implementing scalable industrial AI applications4It is possible to uncover AIs untapped potential with a holistic approach.Digital
103、 and AI can power a new era for continuous improvement in manufacturing beyond the traditional means of driving productivity,thereby unlocking additional value.Although leading manufacturers have already captured significant benefits from AI applications,some are still trying to get started and are
104、looking for a common framework that paves the way for the deployment of AI in manufacturing with a positive return on investment.This study has demonstrated that it is possible to uncover AIs untapped potential with a holistic approach.The development of AI-based applications starts by laying the gr
105、oundwork and some fundamental steps.With a step-by-step approach and the required toolset,the manufacturing sector can gain new perspectives to overcome the most pressing challenges.To do so,the INC Invention Center has developed the AI Navigator,8 a structured toolset to help companies reach the un
106、tapped potential of AI and identify use cases with the strongest chance of successful implementation.The step-by-step guide presented here(figure 4)based on the AI Navigator and consultations from the initiative community provides an example of how to develop application-specific roadmaps and compan
107、ywide actions,from idea generation through evaluation and feasibility analysis to successful implementation.Unlocking Value from Artificial Intelligence in Manufacturing17InitiationKick-off preparation Phase 1IdeationProblem solution fitPhase 2AssessmentAI readiness levelPhase 3FeasibilityImplementa
108、tion concept checkPhase 4ImplementationProduct-market fitStrategyStrategy and leadership commitmentDataData governance structureWorkforceTeam setting&capability buildingDiscoverDefineEmpathizeConfrontInfrastructureDataProcessMindset/cultureBusiness feasibilityTechnical feasibilityData testingTechnol
109、ogy scoutingCompetence validationUse case pre-selectionAgile approachExplainable modelTestingEvaluationDeploymentUse case prioritizationUse case decisionUse case validationPhase 0A step-by-step guide to implementing AI in manufacturingFIGURE 4Source:Adapted from the work of the INC Invention CenterA
110、I is an emerging discipline intended to create a system that amplifies and expands human abilities.To implement a holistic approach independent from industry specifics,there are three fundamentals:strategy,data and workforce.1.Strategic approach and leadership commitment to cultivate the AI transiti
111、on comprehensively,from business units to production facilities.To successfully translate the strategy into action,leading manufacturers have placed significant efforts on change management by actively communicating the set vision and its benefits,investing in workforce upskilling and infrastructure
112、 upgrades and establishing a digital company culture.For AI governance and data ethics mechanisms to be responsible and explainable,they have to be incorporated into this strategy.2.Data governance will be an indispensable part of the process.While collecting the data,it is necessary to build a hub
113、for data flows to manage data availability,usability,integrity and security.AI applications are only built when the data is available;therefore,the most prominent,easily applicable and transferable data structure needs to be built beforehand.3.Cross-functional teams with multi-disciplinary in-depth
114、knowledge in IT,OT,data,analytics and technology engineering and how to organize these with business acumen.Therefore,it is crucial to build an agile team structure for AI projects that revives the collaboration between the technology team and business professionals.AI solutions require collaboratio
115、n with different skill sets and expertise.Even though external sources are used in the transition,in-house experts need to be upskilled.Phase 0:InitiationUnlocking Value from Artificial Intelligence in Manufacturing18The second phase of the AI Navigator methodology focuses on a much more precise ass
116、essment of the maturity level of the company-specific AI use case preselected in phase 1 in order to establish the current status quo.10 This phase aims to determine whether the use case can be successfully integrated with the existing data set,process perspective,infrastructure and culture/mindset.
117、Subsequently,a rough solution hypothesis is to be defined,whereby the target maturity can be specified.Usually,there is a gap between the current and target maturity levels.To close this gap,well-founded recommendations for action must be formulated.Later implementation can only be made possible aft
118、er all to-dos have been processed and the resulting gap between the maturity levels has been closed.Consequently,based on the assessment results,a statement about the expected implementation effort can be made with a high degree of accuracy.The completion of phase 2 further narrows down or focuses o
119、n the use cases that offer comparatively high added value and fulfil the necessary prerequisites for later implementation.Phase 2:AssessmentThe third phase of the AI Navigator focuses on business and technical feasibility.In addition to ROI estimations,critical elements of this phase include:Data te
120、sting:This ensures that enough structured and labelled data sets of the right quality are available to produce the required results.This provides initial insights into the general feasibility of the use case in preparation for implementation.This is essential to the development of the specific techn
121、ology concept.11 Technology scouting:To check whether the previously established solution hypothesis might be developed or purchased,it is recommended to carry out the technology scouting as a next step.In technology scouting,solutions available on the market or at the research and development stage
122、 are analysed,considering the companys individual tech stack.In the best case,existing solutions can be used directly or at least built upon.Competence analysis:If the company requires or desires in-house development,a well-founded competence analysis of the employees skills is necessary to decide w
123、hich cooperation partners are needed externally and which skills can be covered or built up internally.12 During this phase,the solution hypothesis becomes increasingly concrete.If the resulting implementation concept deviates from the original solution hypothesis,it may be necessary to perform the
124、individual steps in phase 3 again.Phase 3:FeasibilityAI projects have the greatest probability of success when businesses conduct a critical early analysis of potential use cases resulting from a problem-oriented approach.After setting the stage in Phase 0,this first phase of the AI Navigator enable
125、s the early prioritization of use cases based on a problem-solution-fit analysis.To discover a set of potential value-adding use cases,a cross-functional team that can analyse and define the main pain points and thus develop the relevant requirements for the solutions needs to be involved.Close comm
126、unication and collaboration with relevant colleagues and end-users at an early stage help ensure the developed solution meets their needs and allows them to adapt progressively to the introduced change.Ultimately,this will increase the acceptance of the deployed applications use.9Phase 1:Identificat
127、ionThe focus of the fourth phase is to clearly define the implementation roadmap.Industrial AI applications need modification,testing and validation of models with iterations,which takes time.Using agile project management methodologies across an open,collaborative environment,including internal and
128、 external team members,can help streamline the process,ensure the implemented application addresses the need of the end-user,and provide a space for innovation and co-creation.It is not enough to pilot an AI use case to leverage the potential of existing data.Scaling the solutions developed is criti
129、cal to success.For this,it is also Phase 4:ImplementationUnlocking Value from Artificial Intelligence in Manufacturing19necessary to consider the data available and the parameters of related use cases,data access,data governance and security,and the required skills early on.13Not many organizations
130、have experience with AI.It is crucial to ensure employees join the transformation to a data-driven future,eliminate fears and prejudices,and establish a culture open to failure.14 To build and sustain trust in AI,transparent and explainable AI models must be implemented and the domain expert must be
131、 aligned with AI recommendations.Leading the projects with agile sprints makes this progress smoother.The maturity of AI models improves over time,which means it is not necessary to change the model according to the first sprint results,as adding additional data sources and increasing the size of av
132、ailable data yields better results.To accelerate the model training curve collaboratively,“federated learning(an approach to machine learning in which the training data are not managed centrally)”15 can distribute the effort for multiple parties across multiple decentralized edge devices or servers
133、holding local data samples without exchanging them.The systematic approach such as that of the AI Navigator enables an additional benefit in terms of the comparability of different ideas and use cases and alignment of the strategy.While it is often possible to develop many technology-oriented ideas,
134、the challenge lies in prioritizing different topics,even in the same department.Using an approach like the AI Navigator and a well-defined evaluation logic allows for the comparison of different use cases across the four phases to determine how relevant they are and what the potential implementation
135、 effort will be in terms of technology,mindset and culture.In addition to building cross-functional teams and involving the domain expert in the development process,the cultural component of a guide such as the AI Navigator enables the early identification of potential action points.16 Developing th
136、e AI literacy of domain experts by collaborating with academia and capability-building centres without focusing on vendor-specific tools is crucial to building the team.The Turkish Employers Association of Metal Industries(MESS),one of Trkiyes largest and most active employer associations,has built
137、the MEXT Technology and Capability Building Center to help its members and companies leverage digital transformation and technologies.The digital factory uses cutting-edge 5G technology on two production lines a discrete production line(end-to-end connected manufacturing)and a continuous production
138、line(digital twin of integrated steel production)on which it has implemented over 160 use cases.MEXT is a real-world demonstration of the initiation phase of the guide.With a strategy to have a state-of-the-art digital model factory,leadership has prioritized AI-related initiatives and use case deve
139、lopment with a clear vision.The AI Lab for Manufacturing is now being built and jointly works on various AI technology governance initiatives with the World Economic Forums Platform for Shaping the Future of Technology Governance:Artificial Intelligence and Machine Learning and with public authoriti
140、es.MEXT is an openly available space to work with academics,manufacturing companies,technology providers and start-ups.It uses a Maestro smart data layer as a data governance solution to serve as a testbed for new technologies,minimize integration efforts and manage all data flows.This approach has
141、shortened the time to value for AI studies due to the availability of clean data with real-time automated contextualization.To accelerate the pace of the implementation of AI solutions,an agile team was built comprising digital transformation and business strategy professionals,data governance exper
142、ts and AI experts from member companies.Additionally,the in-house team has received training on artificial intelligence(AI)-specific knowledge.MEXT has completed more than 150 smart industry readiness index(SIRI)assessments in the automotive,machinery and equipment producers,steel,textile,cement,che
143、mical and food processing industries.Based on the SIRI insights,industry-specific pain points have been defined with the structured pathway of the AI Navigator.The AI use cases prioritized for implementation are workplace safety,AI-based machine health monitoring and forecasting,quality inspection,p
144、rocess optimization and predictive maintenance use cases.With in-house experts and upstream and downstream partners,additional AI use cases are being added and continuously updated.CASE STUDYPilot implementation at MEXT Technology CenterUnlocking Value from Artificial Intelligence in Manufacturing20
145、ConclusionWith a holistic approach,AI can solve some of the most persistent problems in manufacturing and tap into new opportunities that allow companies to increase their operational performance,drive the sustainability agenda and empower the workforce.While organizational and technological challen
146、ges are still hindering the deployment of AI applications at scale,leading manufacturers have successfully seized the AI-derived potential and implemented a wide range of use cases for health and safety,quality,maintenance,production process,supply chains,and resource and energy management.By using
147、a step-by-step approach such as the one highlighted in this white paper,leaders can identify relevant applications and successfully implement them.Moving forward,the World Economic Forum and the Centre for the Fourth Industrial Revolution Trkiye will continue to work closely with stakeholders across
148、 industries and the Centre for the Fourth Industrial Revolution Network,offering MEXT as a unique testing and collaboration system for businesses to:Collectively pilot next-generation AI applications and unlock the untapped value of AI in manufacturing,building on the capabilities of the technology
149、centre,which has positioned itself as a testbed for industry and technology companies that carry out research and development innovations and proof of concept studies to accelerate inclusive technology adoption;Foster a collaborative approach among a diverse system of industry leaders,technology exp
150、erts and academics to develop the right capabilities needed for AI deployment and digital transformation in manufacturing.Manufacturing companies are invited to engage with the Centre for the Fourth Industrial Revolution Trkiye and the Forums Platform for Shaping the Future of Advanced Manufacturing
151、 and Value Chains and Platform for Shaping the Future of Technology Governance:Artificial Intelligence and Machine Learning to collectively make progress on the AI journey in manufacturing and further scale AI capabilities globally,unlocking value for businesses,workers,society and the environment.U
152、nlocking Value from Artificial Intelligence in Manufacturing21ContributorsLead authors Ece Akn ArmutakTechnology Governance Manager,Centre for the Fourth Industrial Revolution TrkiyeMemia FendriInitiatives and Community Lead,Advanced Manufacturing and Value Chains,World Economic ForumWorld Economic
153、ForumFrancisco BettiHead,Shaping the Future of Advanced Manufacturing&Value Chains;Member of the Executive Committee Felipe BezamatHead of Advanced Manufacturing IndustryKay Firth-ButterfieldHead of Artificial Intelligence and Machine Learning;Member of the Executive CommitteeHubert HalopPlatform Cu
154、rator,Artificial Intelligence and Machine LearningVandana MenonEngagement Lead,Centre for the Fourth Industrial Revolution Network and PartnershipsCentre for the Fourth Industrial Revolution TrkiyeTurkish Employers Association of Metal IndustryTun AcarkanDirector,Technology Governance Efe ErdemExecu
155、tive Director,MEXT Technology Center and Centre for the Fourth Industrial Revolution Trkiye Orkun SeerAdditive Manufacturing ManagerUur resinFellow,Centre for the Fourth Industrial Revolution TrkiyeINC Invention CenterBenny DrescherTechnical Director,Artificial IntelligenceToni DrescherChief Executi
156、ve OfficerPatrick KabasciDirector,Digital Operations Business UnitAnne LoosExecutive Director,Artificial IntelligenceUnlocking Value from Artificial Intelligence in Manufacturing22AcknowledgementsThe World Economic Forum and Centre for the Fourth Industrial Revolution Trkiye thank the following indi
157、viduals for their contributions through interviews,use cases and community discussions.alar Aksu Co-Founder and Chief Executive Officer,Sensemore,TrkiyeSerhat Murat AlagzData Analytics Manager,Tpra,TrkiyeBasmah AlBuhairanManaging Director,Centre for the Fourth Industrial Revolution KSA,Saudi ArabiaM
158、eshal AlmashariHead,Digital Design and Delivery,Saudi Aramco,Saudi ArabiaKhalid Sulaiman AlmudayferGeneral Manager,Obeikan Digital Solutions,Saudi Arabiazlem AltnkIntelligent Technologies Senior Director,Martur Fompak International,Trkiyeirin AltokAI-Data Science Chapter Lead,TOFA,TrkiyeHande Bayrak
159、Project Management Director,Turkish Employers Association of Metal Industry,TrkiyePatrick BrownProgramme Director,Predictronics,United StatesMichelangelo CanzoneriGlobal Head,Group Smart Manufacturing,Merck Group,GermanyFrans CronjeChief Executive Officer,DataProphet,South AfricaGner DemiruralSmart
160、Technologies and Innovation Lead,Ford Otosan,TrkiyeHaldun Dinge Executive Director,Digital Production Techniques,Arelik Global,Trkiyeakan EkiciCo-Chief Executive Officer,Khenda,USASercan EsenCo-Founder and Chief Executive Officer,Intenseye,USAJuergen GrotepassChief Strategy Officer Manufacturing,Hua
161、wei,GermanyUur GkdereR&D and Software Development Consultant,Turkish Employers Association of Metal Industry,TrkiyePramod GuptaVice-President,GEP Worldwide,USAJay LeeProfessor and Founding Director,Industrial AI Center,University of Cincinnati,USAHumera MalikChief Executive Officer,Canvass AI,Canada
162、Torbjrn Netland Chair of Production and Operations Management,ETH ZurichJun NiChief Manufacturing Officer,Contemporary Amperex Technology(CATL),Peoples Republic of ChinaYasemin OralCorporate Communications Director,Turkish Employers Association of Metal Industry,TrkiyePamir zbay Head,Operations,Fero
163、 Labs,United StatesErsin ztrkVice-President,Logistics,Digital Transformation,Bosch,TrkiyeJulian SenonerChief Executive Officer,EthonAI,SwitzerlandVolkan Seyok Technical Manager,KARSU Tekstil,TrkiyeAnubhav SinghVice-President,Data and Analytics,Global Supply Chain,Schneider Electric,India Murat Solma
164、zCo-Founder and Chief Operations Officer,Maestro,TrkiyeDaniel SzaboChief Executive Officer,Koerber Digital,GermanyYasir TunerCo-Founder and Chief Executive Officer,Maestro,TrkiyeUnlocking Value from Artificial Intelligence in Manufacturing23Metin TrkayFounder and Chairperson,SMARTOPT,TrkiyeHaydar Vu
165、ralAgile Tribe Lead,Data Science and AI Lead,Tofa,TrkiyeAndre YoonChief Executive Officer,MakinaRocks,Republic of KoreaThe World Economic Forum and Centre for the Fourth Industrial Revolution Trkiye would also like to thank the following individuals for their contributions to the initiative and repo
166、rt.Nilsu AtlanJunior AI Project Specialist,Turkish Employers Association of Metal Industry,TrkiyeJayant NarayanProject Lead,Artificial Intelligence and Machine Learning,World Economic Forum Unlocking Value from Artificial Intelligence in Manufacturing24Endnotes1.McKinsey Global Institute,Notes From
167、The AI Frontier:Insights From Hundreds of Use Cases,Discussion Paper,April 2018,https:/ at the Heart of AI”,The XXXIII ISPIM Innovation Conference“Innovating in a Digital World”,held in Copenhagen,Denmark,June 2022.3.MIT Sloan Management Review,Expanding AIs Impact with Organizational Learning,Octob
168、er 2020,https:/sloanreview.mit.edu/projects/expanding-ais-impact-with-organizational-learning/4.Lee,J.,“Industrial AI:Applications with Sustainable Performance”,Springer and Shanghai Jiao Tong University Press,2020.5.McKinsey Global Institute,Notes From The AI Frontier:Insights From Hundreds of Use
169、Cases,Discussion Paper,April 2018,https:/ Platinion,Artificial Intelligence:Choosing the right approach to machine learning for your needs,https:/ Economic Forum,The Data-Driven Journey Towards Manufacturing Excellence,January 2022,https:/www.weforum.org/whitepapers/the-data-driven-journey-towards-m
170、anufacturing-excellence8.Loos,A.,Sisejkovic,A.,Drescher,B.,“Humans at the Heart of AI”,The XXXIII ISPIM Innovation Conference“Innovating in a Digital World”,held in Copenhagen,Denmark,June 20229.Ibid.10.Ibid.11.Ibid.12.Ibid.13.Ibid.14.Ibid.15.Ludwig,H.,Baracaldo,M.,“Federated Learning:A Comprehensiv
171、e Overview of Methods and Applications”,Springer,202216.Loos,A.,Sisejkovic,A.,Drescher,B.,“Humans at the Heart of AI”,The XXXIII ISPIM Innovation Conference“Innovating in a Digital World”,held in Copenhagen,Denmark,June 2022Unlocking Value from Artificial Intelligence in Manufacturing25World Economi
172、c Forum9193 route de la CapiteCH-1223 Cologny/GenevaSwitzerland Tel.:+41(0)22 869 1212Fax:+41(0)22 786 2744contactweforum.orgwww.weforum.orgThe World Economic Forum,committed to improving the state of the world,is the International Organization for Public-Private Cooperation.The Forum engages the foremost political,business and other leaders of society to shape global,regional and industry agendas.Unlocking Value from Artificial Intelligence in Manufacturing26