Transcription
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review PerspectiveOpportunities and Challenges for AI in AgricultureSupply Chain: A Location Based Review PerspectiveMike DavisCalifornia Polytechnic State University, San Luis Obispo, California, USAAhmed M. Deif *California Polytechnic State University, San Luis Obispo, California, USAThe purpose of this paper is to examine the use of artificial intelligence (AI) in the agriculturalsupply chain (AgSC) and understand the different opportunities and challenges through the lensof an upstream-midstream-downstream location perspective. Through a literature reviewapproach, AI applications in four AgSC domains, namely: operation, decision-making, riskmanagement, and sustainability, along the AgSC are captured. A comparative analysis of thereviewed literature was conducted between the location, technology and business domains of theAgSC. The results captured the dynamics of how AI technologies are evolving and spreading fromupstream to downstream and vice versa. In addition, the analysis led to some recommendations onhow to draft the future research in this field, the investment map, the skill development plan aswell as the sustainability agenda for AgSC.* Corresponding Author. Email address: adeif@calpoly.eduOne industry where artificialintelligence is not gaining traction as quickly,but requires it’s assistance, is the agriculturalsupply chain (AgSC). As much as there is ahigh potential for AgSC to capture AIopportunities, there are many threats for thatindustry that will only grow more severe withthe future. AgSC has many years until itcatches up with its counterparts (compare forexample with apparel supply chain), whichleads to the crux of this papers’ researchobjective.In this paper we are trying, through aliterature review approach, to betterunderstand the AI opportunities andchallenges in AgSC from a locationperspective. The main reason behind such astand is that there are many research effortsattempting to answer the what, why and howI. BACKGROUNDArtificial Intelligence (AI) has grownto become a widely accepted method forstreamlining protocols and automatingprograms to create an efficient and effectivemeans for conveying and executing on inputsgiven. Due to the rapid scalability ofArtificial Intelligence’s prevalence, AI hasbecome adopted into many supply chainsworldwide. It is undeniable that AI willbecome more and more prevalent in modernday technologies, specifically in the industryof supply chain. The benefits of artificialintelligence can be seen throughout financial,manufacturing, and consumer segments ascosts continue to lower and transportationresponsiveness is quickening.Journal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021105
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review Perspectiveof AI in AgSC but to the authors’ knowledge,none had captured the where question. Bysupply chain location, we refer to upstream,midstream and downstream activities asdepicted in figure 1. Along this spectrum, weseek to focus on AI specific applications inselected AgSC domains, namely operation,decision-making, risk management, andsustainability. The selection of these domainswas due to both their importance and theirwide coverage of critical AgSC activitieswith potential AI application.FIGURE 1. AGSC LOCATIONS AND THEIR ACTIVITIESII. RESEARCH APPROACHto 16500 papers. The scope was furtherfocused on the four selected AgSC domainsand AI applications using keywords like,“Automation”, “Agriculture 4.0”, “NeuralNetworks”, “Robots”, “Sustainability”,“Machine Learning”, and “Digital Twins”and thus the number of papers was decreasedto approximately 2500. A representativeselection from this final pool that fulfilled atleast two of the previously stated keywordcriteria in figure 2 and captured AIopportunities and challenges in AgSC wasselected, decreasing the number consideredin this study to 44 papers.The research approach adopted in thispaper was based on conducting a literaturereview followed by a descriptive andcomparative analysis to that review. Thereview was not comprehensive; rather,focused on very recent work in the selectedAgSC domains as well as specific AItechnologies as outlined in figure 2.“ProQuest // ABI Inform” and “GoogleScholar” databases were used. Primaryresults for “AI in AgSC” keyword exceeded146K papers. When focusing on recent workin the last three years, this number decreasedJournal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021106
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review PerspectiveFIGURE 2. LITERATURE REVIEW SELECTION CRITERIAdevelopment of digitals and neural networksat a high level (Demestichas, 2020) (Tzachor,2020) (Denis, 2020).Digital Twins, an AI application thatmirrors the physical reality in a virtual worldfor operation planning and control, will berevolutionary in the world of AgSC (Dash,2019) (Sagarna, 2019) (Erkoyuncu, 2020)(Jones, 2020). Procurement and productionwould benefit greatly for upstreamagricultural supply chain activities (Denis,2020). Midstream, throughput, cycle time,and inventory optimization, transportation,warehousing would all improve due to adigital twin predictive and analysis model(Ahmed, 2015) (Toorajipour, 2020) (Denis,2020) (Food Logistics 4.0, 2021). By creatinga virtual space that near perfectly mimics thephysical Ag space, multiple insights can begained (Tzachor, 2020). Downstream, theinformation surrounding POS for finishedgoods would also be able to provide a wealthof data for future forecasting use (Denis,2020). These uses include product mixselection, product introductions, andincreased diversification of products offered(Dash, 2019) (Toorajipour, 2020) (Black,2020).III. AI APPLICATIONS IN SELECTEDAGSC DOMAINSIn this section, we summarize some examplesof the AI opportunities (through the reviewfindings) as it relates to the selected fourAgSC domains. This will be followed by asummary of some AI applications challengesalong the same domains3.1. AI application Opportunities3.1.1. Operations domain:There are many aspects of AItechnologies that could improve theefficiency of the AgSC operations. Manypapers brought up the applications of AI intraceability (Lakkakula, 2020) (Erkoyuncu,2020), transparency, and efficiency (Trunk,2020) (Walker, 2020). Accurate locationrecordings of products or stages in the supplychain would lower operational errors andincrease cycle times (Denis, 2020).Furthermore, multiple human procedurescould be automated using AI technology(Black, 2020). Data collection, transmission,and analysis (Calatayud, 2019) (Peppes,2020) all can be expedited through theJournal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021107
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review Perspectiveused in future planting decisions. AI can alsoanalyze what choices employees have madein the past which allows management todetermine the best coaching strategies. Thisaction-reaction connection gives themanagement and employee a symbioticrelationship (Bălan, 2019), where both profitoff this combined knowledge.One of the most overlooked, yet mostimportant aspects of AI application in AgSCis the avoidance of technological lock in(Carolan, 2020). The agricultural sector forthe most part has been ingrained in certainideals and methodologies for decades, manyof which have been of detriment to thecompanies within the space (Black, 2020). AIcan be used as a tool to assist with theirtransition out of 20th century technologies. AIcan be employed throughout the entire supplychain, easing the uncertainty in the decisionmaking burden and reducing bullwhip effect(Vaio, 2020). With AI handling a greatnumber of menial, trivial, and automaticdecisions, executives will be able to put theirtime and energy into bigger picture thinkingand taking the AgSC business to the nextlevel.RegardingtheAIselectedtechnologies, digital twins would beremarkably useful for the planning andprediction stages of agricultural development(Dash, 2019) (Jones, 2020). By enteringrelevant historical data and creating a virtualreality, farmers can better account forchanges in weather patterns (Calatayud,2019) (Food Logistics 4.0, 2021) or pestoutbreaks in crops (Marani, 2021). ANNswould also be useful, as a functioning ANNcould help in decisions that would seamlesslyautomate processes upstream and midstreamof the supply chain (Peppes, 2020) (Fountas,2020). The data collected by retailersdownstream can additionally provide insightsinto demand forecasting decisions (Black,2020) and dynamic pricing decisions (Dash,2019) (Food Logistics 4.0, 2021).Artificial Neural Networks (ANN)was shown to be beneficial to the AgSCthrough the speed at which massive amountsof information intake (Vaio, 2020) (Alzoubi,2017) can be made available (An executive’sguide to AI, 2018). Convolutional andrecurrent neural networks have manyapplications such as identifying brand logosduring processing and distribution operationsand make the reconciliation process moreeffective, thereby reducing operation delaycosts (Lakkakula, 2020) (Dash, 2019)(Toorajipour, 2020) (An executive’s guide toAI, 2018) (Food Logistics 4.0, 2021)(Alzoubi, 2017). Farms’ yield can beenhanced, through use of AI based solutionsthat improve process efficiency, reducewastes, and increase the availability ofinformationforfarmingdecisions(Lakkakula, 2020) (Toorajipour, 2020)(Alzoubi, 2017).3.1.2. Decision making domain:Although it is related to the operationdomain, however, AI allows AgSCmanagement to go beyond operation domainand better understand the whole supply chaindynamics offering system wide insights(Ryan, 2020) that could not have been seenwithout it. Utilizing AI in conjunction withwell-designed KPI dashboard will offerdecision makers the capability to capture data(Dash, 2019) (Sagarna, 2019) real time(Toorajipour, 2020) and act accordingly. Forinstance, a farmer could track in real time theprogression of a crop infection and reactaccordingly. Decision makers can also trackthe impact of their previous decisions (Trunk,2020) (Davarzani, 2020) and then use thesedata to adjust and improve their course ofaction from upstream to downstream (Dash,2019) (An executive’s guide to AI, 2018)(Alzoubi, 2017) (Duvniak, 2020). Theimplementation of a new fertilizer on certainareas of fields can provide valuable feedbackJournal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021108
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review Perspectiveonly in regard to defense systems, but AI canbe applied to negotiations, ensuring allstakeholders are in possession of accuratecontracts (Dash, 2019) (Vaio, 2020) (Eteris,2020). A neural network solution coulddetect a falsified contract, code, or user, andtake corrective actions as programmed tostop the threat (Demestichas, 2020)(Toorajipour, 2020). Validation, not trust,will be the defining factor in agriculturalcontracts moving forward.3.1.3. Risk management domain:Regarding risk management in AgSC,AI processes can play a vital role to capture,assess and mitigate multiple risks. Forexample, tracking and tracing diseaseoutbreaks due to food contamination(Demestichas, 2020) can be done seamlesslywith AI technologies. When such an outbreakhappens, the contaminated food would needto have passed through all the distributionfacilities and supplier warehouses (Black,2020). AI can conduct thorough ingredientanalyses during these stages to automaticallydetect diseased products (Tzachor, 2020),thereby reducing the odds of the product everreaching consumption.The agility and resilience of AgSCcan be improved using digital twintechnologies. With what the AgSC chainexperienced for example lately due to theCOVID19 pandemic, such resilience isbecoming an inevitable future strategy for allAgSC (Food Logistics 4.0, 2021). To set abasis for such resilience, digitization and AIbased solutions (like digital twins and ANNmodels) are required for all AgSC fromupstream to downstream (Denis, 2020)(Jones, 2020).Recently, cyberattacks on AgSC havegrown faster than predicted (Food Logistics4.0, 2021). An organization with maliciousintent can effortlessly take control of anation's food supply (Ryamarczyk, 2020)(Liu, 2020) if they tap into weaknesses withina supplier’s IoT system for example. AIsolutions can defend against this threat byusing machine learning and AI authenticationprocesses to correctly validate user access(Peppes, 2020) (Eteris, 2020) (Liu, 2020).Support vector machines, self-organizingmaps, and neural networks models can beemployed to identify and facilitate defensefunctions against antagonistic threats to thesupply chain structure (Trunk, 2020). Not3.1.4. Sustainabledomain:businessmodelsSustainable Business Models (SBMs)have long been sought after as an AgSCbusiness competitive advantage. The triplebottom line framework of sustainability:economic, environmental, and social are wellpresent in AI applications in AgSC, withenvironmental and social often under themost scrutiny from agricultural advocates(Vaio, 2020). For example, to make smarterdecisionsfromtheenvironmentalperspective, executives must first know whatquestions to ask and what data sets toprioritize(Perry,2020).ArtificialIntelligence offers a way to track data andprovide insight (Calatayud, 2019) (Peppes,2020) that can better guide executivestowards a more sustainable method ofoperation (Jones, 2020) (Trunk, 2020).The literature points to many waysthat AI can be used to drive a company’ssustainable development goals. Firstly,analysts can examine past data collectionsfrom their supply chain to route trucking andshipping paths (Ahmed, 2015) (Dash, 2019)in the most efficient way possible (FoodLogistics 4.0, 2021) (Davarzani, 2020).Fertilizers, pesticides, and individualizedirrigation schemes can be utilized throughoutdifferent times to radically lower waterconsumption in farming (Vaio, 2020)(Davarzani, 2020). Ventilation and moistureJournal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021109
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review Perspectivemanagement are another two major issuesthat can be resolved with the implementationof an AI based system (Tzachor, 2020) (FoodLogistics 4.0, 2021). Distribution warehousescan also be completely overhauled usingmachine learning algorithms to completelyautomate the receiving and shipping ofvarious products (Carolan, 2020) (Kopteva,2019). Personnel contentment increases withthe improved transparency that AI brings(Bell, 2020). Emissions can be radicallylowered through a series of neural networks,calculating and reporting the most efficientroutes for shipping equipment and product(Toorajipour, 2020) (Vaio, 2020) (Anandan,2020).On the social level, some worksuggested that AI can also be used to create asmarter and more inclusive organizationalstructure in the agricultural sector.Communication between all levels of anorganization can be vertically integrated intoa company’s flow with AI through vendormanaged inventories, efficient consumerresponse, collaborative planning, forecastingand replenishment, or decentralizedenterprise resource planning systems (FoodLogistics 4.0, 2021) (Black, 2020). This canlead to smarter hiring processes, orderfulfillments, and flexibility in distribution(Jones, 2020) (Kriebitz, 2020) (Black, 2020)(Ryamarczyk, 2020). In the case of aworldwide disaster, in which traditionalAgSC supply channels from farm to client areinterrupted, AI can equip executives with aportfolio of solutions (Walker, 2020), therebyincreasing the chance of a successful pivot.Agile AgSC frameworks have grown inpopularity and can certainly be enhanced byAI implementation (Calatayud, 2019)(Toorajipour, 2020) (Sagarna, 2019). Also,the combination of AI and Agile methods inthe context of prototyping of new agriculturalgenes that are more environmentally friendlycan provide results with speed and frugality(Denis, 2020).3.2. AI application Challenges3.2.1. Costs and Investment Uncertaintydomain:The AI opportunities describedearlier come with some challenges in theAgSC, both in terms of the implementationand the technology behind it. Costsassociated with developing AI solutions likein the case of digital twin or a neural networkexpand outside of the technology itself andwould come on the top of this challenges list(Vaio, 2020). While those costs are high, thecost of training staff, technicians, andexecutives on the usage, implementation, andinterpretation of AI systems is large as well(Food Logistics 4.0, 2021) (Ryamarczyk,2020) (Eteris, 2020). Educational courses andpractices will need to be invested in so thatemployees of experience can developfunctioning, automatic digital systems(Eteris, 2020). The exact economic returnsare uncertain in the majority of cases,resulting in executive buy-in as one of theonly impactful drivers of AI investments(Ryamarczyk, 2020).3.2.2 Creation of potential vulnerabilitiesdomain:As mentioned earlier, AI solutions inAgSC can greatly reduce the risk of securitythreats by creating predictive andpreventative models to account forantagonistic threats. However, if a securitythreat were to gain access to a neural networkor a key AI system, the fallout could beimmense (Peppes, 2020) (Liu, 2020). Thetheme of heavy crop consolidation (Peppes,2020), which has become more common inrelation with Agriculture 4.0, leaves manyvertically integrated supply chains vulnerableto a cyberattack. Blockchain and IoT can pairwith AI as added security levels, but thereJournal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021110
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review Perspectiveapproaches) as well as offering multipleinsights regarding AI application in AgSCfrom mainly a location perspective followedby technology perspective.can’t be any guarantees when connecting theentire supply chain into a single network. Ifan attack were to be successful, it is stillundecided how to identify the responsibleparty when data is distributed so broadly.With so many stakeholders involved in aneural network, digital twin, or AI system ingeneral, the concept of data ownership andprivacy has yet to be unanimously defined(Demestichas, 2020) (Ryamarczyk, 2020).3.2.3. Distinguishingdomain:data4.1. An AgSC Location Perspective:An overview of AgSC location focusdistribution of reviewed papers is displayedin Figure 3. The majority of the papersfocused on AI applications at the upstreamsector of the AgSC (41.8%) while 29.7% ofpapers focused on the midstream and 28.6%of the papers focused on the downstream AIapplications.Table 1 depicts the covered literaturereview work for different AI applicationswithin the selected four AgSC domains alongthe AgSC locations.relevancyFor AI to be successfully deployedinto a system, there must be precise andrelevant historical data (Trunk, 2020)(Davarzani, 2020). While precise data ispresent in many large organizations, decidingwhat is and is not relevant can only be foundthrough numerous trials by engrainedindividuals. The difference in data used in anAI system can determine whether a crop’syield is average or exceptional. It is difficultto mix hard and soft skills/variables into asingle program as well (Bălan, 2019). Anoveremphasis on one or the other could leadto wildly inaccurate results. This can beespecially difficult in the agricultural sector,as tradition is emphasized heavily incomparison to other industries. Ensuring thatthe data is properly structured by a qualifiedstakeholder is necessary for a validimplementation (Kopteva, 2019) (Eteris,2020).IV. REVIEW ANALYSISAPPLICATION IN AGSCOFTable 1 conforms that in general, thesheer weight of research focused on theupstream processes. Furthermore, operationmanagement domain in that upstream focusis shown to have the largest AI application inAgSC, followed closely by the decisionmaking domain (especially upstream).Sustainable business models had the leastresearch focus in the context of this review,pointing towards clear opportunities forfuture research in this domain.In Tables 2, 3, and 4, we list theopportunities and challenges offered by thespecific AI applications considered in thisreview (digital twin, machine learning,robotic solutions and neural network models)along AgSC upstream, midstream anddownstream respectively. An opportunity isviewed as a potential benefit from theapplication and a challenge is anything ptance.AIIn this section, the outlined reviewoutcomes are further analyzed using multiplecomparative approaches. The comparativeanalysis aims at summarizing the capturedreview (using tabular and figurativeJournal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021111
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review PerspectiveFIGURE 3. DISTRIBUTION OF REVIEWED PAPERS BY AGSC LOCATIONTABLE 1: IMPROVING THE AG SUPPLY CHAIN USING AI-BASED SOLUTIONS DOMAIN CATEGORIZATION BY streamAutomated tooling equipment insmart farming (Carolan, 2020)(Kopteva, 2019)Plant gene predictive models(Carolan, 2020)Transparency, traceability, andefficiency (Lakkakula, 2020)(Erkoyuncu, 2020) (Trunk, 2020)(Walker, 2020) (Black, 2020) (Liu,2020)Faster data collection, transmission,and analysis (Calatayud, 2019)(Peppes, 2020)Automatic irrigation systems basedon environmental factors (Peppes,2020) (Shi, 2019) (ZounematKermani, 2020)Expedite Procurement and Production(Denis, 2020) (Davarzani, 2020))(Ryan, 2020)Opportunities for rapid scalability(Denis, 2020)Data mapping and simulations usingDigital Twin (Erkoyuncu, 2020)Maximize asset utilization (Bell,2020)DES to estimate on handinventory levels (Ahmed,2015)Reduces TransportationCosts using ANNs(Ahmed, 2015) (Dash,2019) (Food logistics 4.0,2021) (Davarzani, 2020))Lowers delay costs andquickens thereconciliation process(Lakkakula, 2020) (Foodlogistics 4.0, 2021)(Fountas, 2020)Machine HealthMonitoring andpredictive/preventativemaintenance (Dash, 2019)(Jones, 2020)Efficient lot sizing tomaximize utilization ofresources (Toorajipour,2020) (Ryan, 2020)Detection of defectiveproducts using CNNs (Anexecutive’s guide to AI,Customizationenhancement and resourcecoherence (Bourke, 2019)(Ryan, 2020)Reduce inventory holdingcosts to increase inventorysize, leading to increasesprofits (Ahmed, 2015)(Food logistics 4.0, 2021)Reactions in Agile andLean methodologies(Calatayud, 2019)(Toorajipour, 2020)(Sagarna, 2019)Assists in productintroductions (Dash,2019)Product Mix / Lineoptimization (Toorajipour,2020)Output of adaptivefeedback reports usingRNNs (An executive’sguide to AI, 2018) (Foodlogistics 4.0, 2021)Journal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021112
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review PerspectiveDecisionMakingAutomatic harvesting protocols(Anandan, 2020) (Shi, 2019)Precision sensors to for automatedguided vehicles (Anandan, 2020)(Food logistics 4.0, 2021)(Rymarczyk, 2020) (Marani, 2021)Improved working conditions foremployees on site (Food logistics 4.0,2021)Efficient crop cultivation through soilanalyses (Majumdar, 2019)(Milunović, 2018)Increased yields (Fountas, 2020)(Ryan, 2020) (Marani, 2021)Remote Sensing to track crop yields(Fountas, 2020) (Ryan, 2020)(Marani, 2021)Rapid Crop identification (Perry,2020)Tracking and reduction of waterusage (Zounemat-Kermani, 2020)(Duvniak, 2020)Detection of crop color to identifyripeness (Marani, 2021)The agricultural drone for fieldassessment (Carolan, 2020) (Shi,2019)2018) (Food logistics 4.0,2021)Detect company logos orcodes for sorting (Anexecutive’s guide to AI,2018)Throughput, Cycle time,and Inventoryoptimization (Denis,2020)Enhance flexibility (Jones,2020) (Black, 2020)(Rymarczyk, 2020)Reduce downtime (Bell,2020)Supply chain networkresilience (Black, 2020)Lower levels ofoperational/capacity slack(Black, 2020)Efficient POS for finishedgoods (Denis, 2020)Model Based PredictiveControls, Future EarningsModeling using DigitalTwins (Jones, 2020)Upgrade effectiveness offleet management (Bell,2020)Increase volume, variety,and velocity of productsold (Food logistics 4.0,2021)Greater diversification inproducts that can be sold(Black, 2020)Comparisons ofcost/benefit plans usingAI (Duvniak, 2020)Avoidance of technical / knowledgelock-in (Carolan, 2020)Decision support system for farm andcrop management qualitativetraceability (Tzachor, 2020) (Peppes,2020)Assists with data accuracy andoutputs (Demestichas, 2020)Utilizes real-time KPIs forinstantaneous feedback (Dash, 2019)(Sagarna, 2019)Sift through unambiguous, complexdata to drive smart reporting methods(Bell, 2020) (Duvniak, 2020)Improve organization structures(Trunk, 2020)AI can augment human capabilitiesand vice versa (Trunk, 2020)Prediction of water and irrigationrevenues and costs (ZounematKermani, 2020) (Duvniak, 2020)Agricultural technology providers canprovide insights using big data (Ryan,2020)Can address stressors in farmers(Beseler, 2020)Creates data chainsdecision makers canfollow (Carolan, 2020)DSNs to facilitateproduction planning thatmatches market demand(Calatayud, 2019) (Trunk,2020)Can enable a symbioticrelationship betweenmanager and employee(Bălan, 2019)Enables accountabilitywhen working withmultiple stakeholders(Dash, 2019)Language translations forMultinationalCorporations (Anexecutive’s guide to AI,2018)Collaborative Planning,Forecasting andReplenishment (Foodlogistics 4.0, 2021)Automation of commoditytrading (Lakkakula, 2020)(Black, 2020)Provide information bankfor executives to balancehard and soft objectives(Bălan, 2019)Auto-generatingsuggestions to clientsbased on previous orders(Bălan, 2019)Provide information onclient consumptionpatterns and perceptions(Dash, 2019) (Anexecutive’s guide to AI,2018) (Alzoubi, 2017)(Duvniak, 2020)Increase demand forecastaccuracy (Dash, 2019)(Food logistics 4.0, 2021)(Black, 2020)Implement dynamicpricing (Dash, 2019)(Food logistics 4.0, 2021)Journal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021113
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review PerspectiveCreates a strong portfolio of solutions(Walker, 2020)Leaner logistics planningthrough decentralizeddecision making (Foodlogistics 4.0, 2021)RiskManagementTraceability of food production(Calatayud, 2019) (Peppes, 2020)(Shi, 2019)Response to changing weatherpatterns (Calatayud, 2019) (Foodlogistics 4.0, 2021)Simulating and evaluating thedegradation of the biophysicalenvironment (Tzachor, 2020)Simulating future yield performances(Tzachor, 2020)Properly store and manage datavulnerabilities from security threats(Peppes, 2020) (Liu, 2020)Tracing of water supply quality(Duvniak, 2020)Distributed Ledgers forexpedited contracts(Demestichas, 2020)Pinpoint where assetmaintenance is needed incase of a breakdown(Calatayud, 2019)Ensure data integrity fromtheft (Peppes, 2020)Increase ability to pivotquickly regarding naturaldisasters (Black, 2020)Tracking and internal /external tracing(Demestichas, 2020)(Tzachor, 2020) (Peppes,2020)Ingredient analysis fordiseases (Demestichas,2020)Detecting real timedisease outbreaks(Tzachor, 2020)Assess the likelihood thata transaction is fraudulent(An executive’s guide toAI, 2018)SustainableBusinessModelsAutomatic food sorting (Vaio, 2020)Insurance of hygiene standards andpractices (Vaio, 2020)Optimize fertilizers, pesticides andsystemized irrigation (Vaio, 2020)(Davarzani, 2020)Ventilation and moisturemanagement (Tzachor, 2020) (Foodlogistics 4.0, 2021)Improve marketcorrespondence (Vaio,2020) (Bell, 2020) (Foodlogistics 4.0, 2021)Reduce overall emissions(Toorajipour, 2020)(Vaio, 2020)(Anandan,2020)Can balance betweeneconomic, environmental,and social priorities (Vaio,2020)Create effective SDGs(Vaio, 2020)(Kriebitz,2020)Reduce the cost of foodrecalls (Lakkakula, 2020)AI literate executives willfacilitate an AI centeredsupply chain (Bălan,2019) (Trunk, 2020)TABLE 2: OPPORTUNITIES AND CHALLENGES OF AI APPLICATIONS INUPSTREAM tal TwinSweeping cost reductions (Jones, 2020)Reduce risk, design & reconfiguration times(Jones, 2020)Maintenance decision making (Jones, 2020)Modularity to fit different environments(Rymarczyk, 2020)Ability to Imagine, Design, Realize, Support,Retire (Jones, 2020)Digital Twins are heavily specified, nogeneral software or standard (Jones, 2020)Understanding what level of fidelity isdifficult to define (Jones, 2020)Training of existing staff may be costly(Food logistics 4.0, 2021)Journal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021114
Mike Davis and Ahmed M. DeifOpportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review PerspectivePreset agile frameworks (Denis, 2020) (Sagarna,2019)Efficient Prototyping (Rymarczyk, 2020)MachineLearningUtilizing Chat-bots to educate rural communities(Ekanayake, 2020)Fruit detection and counting (Marani, 2021)Able to increase shelf life of products (Foodlogistics 4.0, 2021)Rule based expert systems for manufacturingcontracts (Toorajipour, 2020)Reduction in delay costs and the reconciliationprocess (Lakkakula, 2020)Predict and analyze the mental health of farmworkers
Opportunities and Challenges for AI in Agriculture Supply Chain: A Location Based Review Perspective Journal of Supply Chain and Operations Management, Volume 19, Number 2, December 2021 108 Artificial Neural Networks (ANN) was shown to be beneficial to the AgSC through the speed at which massive amounts
