AI Report2

International Journal of Production Research
ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/tprs20
Artificial intelligence in supply chain and

operations management: a multiple case study
research
Violetta Giada Cannas, Maria Pia Ciano, Mattia Saltalamacchia & Raffaele
Secchi
To cite this article: Violetta Giada Cannas, Maria Pia Ciano, Mattia Saltalamacchia & Raffaele
Secchi (12 Jul 2023): Artificial intelligence in supply chain and operations management:
a multiple case study research, International Journal of Production Research, DOI:
10.1080/00207543.2023.2232050
To link to this article: https://doi.org/10.1080/00207543.2023.2232050
© 2023 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group.
Published online: 12 Jul 2023.
Submit your article to this journal
Article views: 12008
View related articles
View Crossmark data
Citing articles: 4 View citing articles
Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tprs20
INTERNATIONAL JOURNAL OF PRODUCTION RESEARCH
https://doi.org/10.1080/00207543.2023.2232050
Artificial intelligence in supply chain and operations management: a multiple

case study research
Violetta Giada Cannasa , Maria Pia Cianob , Mattia Saltalamacchiac and Raffaele Secchic
a School of Industrial Engineering, Carlo Cattaneo – LIUC University, Castellanza, Italy; b Nottingham University Business School, University of
Nottingham, Nottingham, UK; c School of Economics and Management, Carlo Cattaneo – LIUC University, Castellanza, Italy
ABSTRACT ARTICLE HISTORY

Artificial intelligence (AI) is increasingly considered a source of competitive advantage in opera- Received 8 December 2022
tions and supply chain management (OSCM). However, many organisations still struggle to adopt Accepted 22 June 2023
it successfully and empirical studies providing clear indications are scarce in the literature. This KEYWORDS
research aims to shed light on how AI applications can support OSCM processes and to identify Artificial intelligence;
benefits and barriers to their implementation. To this end, it conducts a multiple case study with operations management;
semi-structured interviews in six companies, totalling 17 implementation cases. The Supply Chain supply chain management;
Operations Reference (SCOR) model guided the entire study and the analysis of the results by target- SCOR; industry 4.0;
ing specific processes. The results highlighted how AI methods in OSCM can increase the companies’ multiple-case study
competitiveness by reducing costs and lead times and improving service levels, quality, safety, and
sustainability. However, they also identify barriers in the implementation of AI, such as ensuring data
quality, lack of specific skills, need for high investments, lack of clarity on economic benefits and lack
of experience in cost analysis for AI projects. Although the nature of the study is not suitable for wide
generalisation, it offers clear guidance for practitioners facing AI dilemmas in specific SCOR processes
and provides the basis for further future research.
1. Introduction (Wong et al. 2022), resilience (Belhadi et al. 2022), and

In recent years, artificial intelligence (AI) has assumed sustainability (Olan et al. 2022; Pournader et al. 2021).
a prominent role in the Operations and Supply Chain The common goals of these applications are reducing
Management (OSCM) literature. Indeed, AI has proven the time required for decision support, reducing human
to be of great value to companies and is considered a resources for repetitive tasks, and increasing capacity
key technology for Industry 4.0. It owes its popularity utilisation. (Helo and Hao 2021).
to its ability to make OSCM processes responsive to the AI is commonly defined as a field of computer science
various challenges that characterise today’s world, such that encompasses the development of systems that can
as unpredictable disruptions, dynamic customer expec- perceive and interact with the environment in the form
tations, intense global competition, increasingly strong of text, video, audio, and more through approaches such
and urgent pressure to digitalise companies, and ever- as speech, vision, and natural language processing (NLP);
evolving technological innovations. (Fosso Wamba et al. that can learn from the experience provided by historical
2021; Helo and Hao 2021; Zeba et al. 2020). data through machine learning (ML) methods; and that
AI can be applied in several OSCM processes, such can make decisions that normally require human intel-
as order intake, supplier selection, quality control, pro- ligence through approaches such as planning, optimisa-
duction planning and control, smart connected prod- tion, simulation, modelling, programming, and expert
ucts, services and maintenance, transportation, ware- systems (Helo and Hao 2021; Pournader et al. 2021).
house management, sales processes, and customer inter- Among the AI methods, ML techniques are con-
faces (see Table 1; Helo and Hao 2021). Furthermore, sidered the most applied in OSCM, and their applica-
AI applications can support specific purposes, such as tions have recently gained interest among researchers
inventory management (Svoboda and Minner 2022), because of their capability to rapidly and intelligently
demand forecasting (Kantasa-ard et al. 2021), risk man- manage big data and handle nonlinear problems that are
agement (Baryannis et al. 2019; Wong et al. 2022), agility widespread in real-world supply chains (Riahi et al. 2021;
CONTACT Maria Pia Ciano [email protected] Wollaton Rd, Lenton, Nottingham NG8 1BB, UK
© 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the Accepted
Manuscript in a repository by the author(s) or with their consent.
2 V. G. CANNAS ET AL.
Tirkolaee et al. 2021). This supports decision-making by The remainder of the article is organised as follows.
exploiting both historical and real-time data with only Section 2 reports a literature review regarding the use of
minor human involvement (Zeba et al. 2020), improves AI in OSCM processes organised according to the supply
performance (Fosso Wamba et al. 2021), and enhances chain operations reference (SCOR) model, which is used
competitive advantage (Dubey et al. 2020). as a reference model throughout the article. Section 3
Notwithstanding the relevance of AI for OSCM, there describes the methodology by which this research was
is still little research on the role of AI in this field, conducted, that is, multiple case studies, and outlines
and expertise on this subject is still limited (Helo and the cases considered and the steps followed for data col-
Hao 2021). As a result, nowadays, companies struggle to lection and analysis. The results and analysis are pre-
implement AI in their OSCM processes and face numer- sented in Section 4. Section 5 discusses the research
ous barriers. Many companies have failed to fully com- questions. Finally, the limitations of this study, its impli-
prehend the potential of AI (Pournader et al. 2021). Many cations, and future research directions are discussed in
companies lack a clear strategy, solid technology infras- Section 6.
tructure, and the full commitment of top p management
and are discouraged by the high expenditure required to
2. Literature review
build skills and acquire technology (Dubey et al. 2020;
Helo and Hao 2021). A cause of the lack of clear direc- The review of the existing literature has identified sev-
tion, trust, and motivation for a conscious transition to eral applications of AI methods in OSCM. The results
AI may be found in the scarcity of empirical studies in the were classified following the SCOR model, as proposed
literature on AI to support OSCM processes, and the lack by other recent studies (Chehbi-Gamoura et al. 2020),
of examples of companies that have already implemented focusing on five core processes: ‘plan’, ‘source’, ‘make’,
AI in OSCM. This certainly precludes obtaining insights ‘deliver’, and ‘return’.
and practical guidance on approaching the subject. More- The survey of the literature consisted of keywords
over, recent literature reviews on AI in OSCM underlined searching through Scopus database (www.scopus.com)
that the number of studies dedicated to the applications of and was conducted in March 2023. Given the broad scope
ML algorithms in managing a supply chain is inadequate of the research, which addresses the application of AI
(Tirkolaee et al. 2021) and highlighted the need to analyse to the five core processes of the SCOR model, we lim-
practical challenges and propose models to fit AI in oper- ited the literature survey to articles that included, in their
ational environments to solve problems (Dhamija and title, keywords that expressly denote the area of inves-
Bag 2020). tigation. Also, we refined the results by applying exclu-
Therefore, the current context demonstrates the need sion criteria that limited the search to English language
to further investigate the applications of AI in OSCM articles, published by peer-reviewed journals, related to
through more experience and insights from in-depth case the subject area of ‘Business, management and account-
studies in industry (Fosso Wamba et al. 2021; Pournader ing’. The final query used for Scopus searching was the
et al. 2021). To fill this gap, the present study aims to following:
answer the following research questions (RQs) by empir- (TITLE (‘artificial intelligence’) OR TITLE (‘AI’) AND
ically analysing multiple case studies: TITLE (‘plan∗ ’) OR TITLE (‘sourc∗ ’) OR TITLE
(‘mak∗ ’) OR TITLE (‘deliver∗ ’) OR TITLE (‘return∗ ’))
RQ1. How do AI applications support supply chain and
AND (LIMIT-TO (SRCTYPE, ‘j’)) AND (LIMIT-TO
operations management processes?
(SUBJAREA, ‘BUSI’)) AND (LIMIT-TO (LANGUAGE,
RQ2. What are the benefits achieved by companies when ‘English’))
using AI to support supply chain and operations man-
agement? Then, from the resulting list of documents, we screened
titles, abstracts, and full text to exclude papers that were
RQ3. What are the barriers tackled by companies when not related to the main scope of this study, and we
implementing AI to support supply chain and operations
management?
conducted ‘snowballing’ through their reference list to
include additional papers that contributed to the main
By answering these questions, this study contributes scope of the study. This procedure resulted in an asso-
to the advancement of theory and practice. Through ciation of the AI methods with the SCOR processes, with
the empirical analysis of case studies, the authors can details of the application of these methods and the ben-
report on real-life applications of AI in support of efits achieved through it. The authors then analysed the
OSCM, identify the benefits obtained through these literature related to the main barriers tackled by compa-
applications, and classify the barriers faced during their nies when implementing AI to support OSCM. The main
implementation. results are summarised in Table 1.
INTERNATIONAL JOURNAL OF PRODUCTION RESEARCH 3
2.1. Plan (Cantini et al. 2022; Priore et al. 2019). Aktepe, Yanık, and
Ersöz (2021) underlined the success of ML applications,
The Plan processes include ‘gathering of requirements,
artificial neural network and support vector regression,
gathering of information on available resources, balancing
for the demand and inventory management of spare parts
requirements and resources to determine planned capa-
in the construction industry, which is considered a com-
bilities and gaps in demand or resources, and identifying
plex matter because of the high number of parts managed
actions to correct these gaps’ (APICS 2017). The litera-
and the risk of stock obsolescence, intermittent demand,
ture investigates AI in planning processes, mostly in the
and high responsiveness required. Lolli et al. (2019) pro-
domains of demand forecasting, demand planning, and
posed ML, support vector machines with a Gaussian ker-
inventory management.
nel and deep neural networks, for forecasting and stock
As far as demand forecasting and planning are con-
control of intermittent demand. Selecting the best inven-
cerned, datasets related to demand, in most real-world
tory management policy, achieving an average accuracy
problems, are nonlinear, complex, and have ambigu-
of 88% and reducing operating costs proved to be a major
ous structures and patterns. This can disturb forecasting
benefit of the proposed methods.
accuracy, which is fundamental for ensuring the quality
of managerial decision-making. To address this issue, the
literature proposes AI methods based on ML as effective 2.2. Source
methods when adequate data are available. In the energy
The source processes include ‘issuing purchase orders,
industry, Tutun et al. (2022) proposed AI for forecasting
scheduling deliveries, receiving orders, validating orders,
electricity consumption through cohort intelligence &
storing goods, and accepting suppliers’ invoices’ (APICS
adaptive neural fuzzy inference system, and Khashei and
2017). The literature investigates the use of AI in source
Chahkoutahi (2021) proposed AI for electricity demand
processes in the domains of supplier selection and verifi-
forecasting through fuzzy hybrid intelligence based sea-
cation of incoming order quality.
sonal models. In these cases, ML methods learn from
Supplier selection is one of the most crucial stages of
data and adapt without requiring much information,
sourcing, because of the important upstream role of sup-
thus ensuring generalisability and flexibility. In the food
pliers in terms of time, cost, and quality. Most studies on
industry, Priyadarshi et al. (2019) show that ML, and in
this topic have used a combination of fuzzy set theory
particular, long short-term memory (LSTM) and support
and multi-criteria decision models (MCDM) for sup-
vector regression (SVR), applied to demand forecasting
plier selection to address the vagueness of information on
at retail stage for selected vegetables increases perfor-
supplier attributes and the conflicting objectives of iden-
mance in terms of inventory turnover and days of stock
tifying suppliers that are both reliable and cost-effective
coverage to prevent potential stockouts and decrease the
(Sharma et al. 2022).
bullwhip effect. The benefits derived from ML methods
According to Riahi et al. (2021), other popular tech-
are the reduction of forecast error and days of coverage,
niques include artificial neural network (ANN) and
growth of the inventory turnover index, and reduction in
genetic algorithms. An interesting aspect of ANN con-
the number of days out of stock. However, the ML meth-
cerns the increasing focus on green supplier selection or
ods have certain limitations. First, they require a large
supplier selection in the health sector (e.g. Bahadori et al.
amount of historical data to reach the desired accuracy;
2020; Fashoto et al. 2016). Most studies adopting artificial
patterns in the underlying data are also not considered;
neural network (ANN) combine several techniques. For
and finally, they do not have a very good performance in
example, Kuo, Wang, and Tien (2010) combined an ANN
dealing with ambiguities and uncertainties (Khashei and
with data envelopment analysis (DEA) and a network
Chahkoutahi 2021). To overcome these limitations, the
analytic process (ANP) to select the best green supplier
use of hybrid methods, such as an adaptive neural-based
and evaluated the model to a case company operating in
fuzzy inference system (ANFIS) and fuzzy seasonal mul-
the electronic industry. Bahadori et al. (2020) proposed
tilayer perceptron (FSMLP), which combine statistical,
a supplier selection model for hospitals to identify the
metaheuristic, and ML approaches, is proposed. Different
best medical equipment supplier using a combination of
applications of these algorithms can be used to forecast
ANN and fuzzy VIKOR. Regarding genetic algorithms,
energy demand (Khashei and Chahkoutahi 2021; Singh
among the many studies, Fallahpour et al. (2017) pro-
and Challa 2016; Tutun et al. 2022).
posed a new genetic-based intelligent approach, gene
In addition, ML determines the best inventory man-
expression programming (GEP), to improve the sup-
agement policy to obtain optimal inventory control and
plier selection process and overcome other AI methods’
supply chain configuration with minimum cost in the
drawback of being a black box system; indeed, the GEP
context of ‘intermittent demand products’, or spare parts
method provides an explicit mathematical model for
supplier performance based on the determined criteria (DL) in machine vision systems, originally using Otsu’s
and can provide a fast evaluation of potential new sup- algorithm in conjunction with the OBISA algorithm, for
pliers. This study evaluated the method through a case identifying product scraps, achieving an accuracy rate of
study in the garment industry by applying three accu- 85.11%.
racy measures criteria, including MSE, RMSE, and MAE,
and found that the GEP based model performs better
2.3. Make
than ANN integrated model. Luan et al. (2019) presented
a hybrid method based on a genetic algorithm and ant The make processes include ‘processing, maintenance,
colony optimisation to combine the advantages of the repair, overhaul, recycling, refurbishment, manufactur-
former, namely, high initial speed-up convergence, and ing and other common types of material-conversion pro-
the merits of the latter, namely, parallelism and effective cesses’ (APICS 2017). The literature investigates the use
feedback. The method was tested via a simulation case of AI in the make processes, regarding scheduling, man-
referring to extant literature, without using specific case ufacturing, quality, and maintenance.
studies. Green supplier selection, as an increasingly topi- The literature shows that ANN-based techniques are
cal theme, is also addressed by methods based on genetic employed in scheduling problems to determine the opti-
algorithms (Hamdan and Jarndal 2017; Yeh and Chuang mal paths of re-entrant products, while the simulated
2011; Zhang and Cui 2019). annealing algorithm finds application in the prediction
According to Tirkolaee et al. (2021), among ML tech- of machine failures in order to reduce repair costs (Kim
niques, the literature on supplier selection problems and Kim 2021; Wu et al. 2020). Fuzzy logic (FL) can be
presents both supervised learning and reinforcement applied to software for defining the best job shop flow,
learning techniques. In the former, the studies focus on allowing a reduction in the production cycle time (Chan,
the combination of decision trees and support vector Prakash, and Mishra 2013; Mohapatra, Benyoucef, and
machines (SVM), and in the latter, on the Q-learning Tiwari 2013). An alternative may be genetic algorithms,
technique. In particular, the authors reported the key also adopted in software for defining the best flow in pro-
study by Guo, Yuan, and Tian (2009) based on poten- duction scheduling problems (Lei 2010). ML can also
tial SVM and decision trees as the main reference for help to better plan customisation needs. For example,
the procedure to follow and for demonstrating that, com- Chen, Tai, and Hung (2012) applied NNs to group similar
pared to traditional MDCM and original SPV methods, customisation needs. They then used existing inventory
it reduces the number of binary classifiers, increases the information to select parts for production, which dra-
accuracy and decreases the computation time in different matically reduced the cost throughout the supply chain
classifications, and selects the most ‘informative’ features compared to a human decision.
to develop classifiers. Instead, Islam, Amin, and Ward- Since the beginnings of smart manufacturing and
ley (2021) adopted a relational regressor chain method Industry 4.0 concepts, the literature linking production
to determine future demand and used it to feed a multi- and AI techniques has focused heavily on the advantage
objective programming model to identify suitable suppli- that real-time data obtained through IoTs and sensors
ers and order quantities from each supplier. can bring for controlled and self-regulated production
Verification of the quality of incoming orders in the (Lăzăroiu, Kliestik, and Novak 2021; Lee et al. 2018). For
literature is addressed by learning techniques. In gen- example, the optimisation of energy and resource util-
eral, these techniques can be applied at all quality con- isation in production can be done using the k-means
trol stages. Mani et al. (2021) considered an incom- algorithm that creates reliable and updated data on their
ing inspection problem as a reinforcement-learning task. real-time monitoring (Manimuthu et al. 2021; Omwando
They developed a temporal difference learning approach et al. 2018). Wuest, Irgens, and Thoben (2014) proposed
to predict the acceptance and rejection paths of raw mate- an approach to increase the process and product quality
rials by analysing different paths that the raw materials in manufacturing by monitoring data based on prod-
could travel. The method was tested by using Origi- uct status and using a combination of SVM and cluster
nal Equipment Manufacturers (OEM) Incoming Inspec- analysis.
tion process as case study. Zhao, Zhang, and Zhang Along with production control and monitoring, there
(2021) applied an ANN-based technique, specifically is the topic of quality. Product testing and fault detection
based on the convolutional neural network (CNN), in in the process or intermediate outputs are fundamental
a visual recognition system for the identification and to avoid jeopardising the quality of the product and the
classification of different types of defects in the tex- safety and efficiency of the process. Many studies have
tile industry, achieving an accuracy rate of 97.8%. Una- focused on visual inspection. For instance, Villalba-Diez
jan, Gerardo, and Medina (2019) applied deep learning et al. (2019) improved the defect detection performance
during the process of producing gravure cylinders, forest, and XGBoost for predicting the demand and a
reduced quality inspection costs by process automation, visual representation of the results of stocking to better
and developed a deep neural network (DNN) soft sensor understand how to avoid excessive overstocks.
that compares a scanned surface to the used engrav- Regarding process enquiry and quote, NLP imple-
ing file, achieving an automated classification accuracy mented in chatbots can be used to answer generic cus-
rate of 98.4%. Also, Sharma (2023) used the digital twin tomer enquiries and even to automate direct sales (Toora-
method to design, simulate, and validate an automated jipour et al. 2021; Zdravković, Panetto, and Weichhart
battery module inspection system that uses cobots and 2022). These virtual assistants can be extremely effec-
machine vision to inspect electric vehicle batteries for tive for the automation of customer service enquiries, as
defects. The simulation results shown that this auto- they can personalise communication with customers by
mated inspection can reduce time and increase accuracy detecting their personalities and adapting to them (Shu-
of inspection, if compared to human checking. manov and Johnson 2021). FL can enable timely pricing
Fault detection opens the topic of maintenance, par- to complete quotes for customers (Leung et al. 2019). This
ticularly predictive maintenance. Predictive maintenance resulted in a reduction of the average planning time of
primarily involves applications of techniques based on offers by 60% and an increase in the acceptance rate of
ML (Carvalho et al. 2019; Kumar and Hati 2021). For quotes by 78%. The artificial bee colony algorithm can be
instance, Luo et al. (2019) developed a deep learn- applied to an automated system that can execute simulta-
ing model to automatically select impulse responses neous quotes by matching the various strategies of sellers.
from vibration signals during long-term operations to In comparison to any other solution, this increases the
enable early fault detection under time-varying condi- number of quotes by up to 99%, which leads to a deal
tions. Encapera, Gosavi, and Murray (2021) described and increases the average buyer’s utility by up to 80%
the use of Reinforcement Learning to solve large-scale (Kolomvatsos et al. 2016).
Markov11 decision processes for total productive main- Literature reports the use of genetic algorithms applied
tenance and quicken optimisation. Kaparthi and Bum- through data mining also in support of shipping, as it can
blauskas (2020) described the application of an algorithm optimise product delivery methods and logistics in gen-
based on a tree model to predict machine failures using eral. This application showed a reduction in distribution
data from the cloud. This results in greater customer costs and delivery times, achieving an average customer
satisfaction as they can thus have a better after-sale main- satisfaction of 98.75% (Li and Xiao 2021).
tenance service. Regarding picking, Nowak, Hewitt, and White (2012)
Despite the several benefits of AI techniques in the proposed the application of tree models to determine
Make processes, Lee et al. (2018) stated that a possible which use of loads could reduce the total cost of trans-
challenge is that AI techniques require massive and clean port, achieving an average reduction of 5%, which could
datasets with minimum biases. For example, in visual reach 8% for the largest loads.
inspection, some processes may be of such a nature that Studies can be found in the literature using ANN-
each new product implies that the DNN model has not based techniques and genetic algorithms for logistics
been previously confronted with these images. There- optimisation. Among the applications of ANN are tools
fore, the training and testing of DNN is a real challenge for the recognition and analysis of identification codes
(Villalba-Diez et al. 2019). and tags in pallets. These tools make it possible to
optimise warehouse layout, improve the storage of raw
materials, reduce procurement times, and reduce trans-
2.4. Deliver
port time and costs (Ellefsen et al. 2019). Moreover, an
The delivery processes include ‘receiving, validating, and ANN-based decision support system can plan the best
creating customer orders, scheduling order deliveries, truck route. This leads to the reduction of workloads
picking, packing and shipping, and invoicing customers’ through adaptive routing of trucks and the reduction
(APICS 2017). The use of AI in delivery processes has of their waiting times (Hill, Jü, and Böse 2016). Other
been investigated in the literature, in the domains of applications can be found in driver assistance systems
stocking, process enquiry and quote, shipping, picking, for freight vehicles. They enable greater efficiency, both
and logistics optimisation. in shipping through the creation of better routes and
Regarding stocking, Priore et al. (2019) adopted in systems through predictive maintenance (Loske and
inductive learning to determine the best procurement Klumpp 2021).
policy, achieving 88% accuracy and reducing operat- Furthermore, Dosdoğru, Boru İpek, and Göçken
ing costs. Sharma and Singh (2021) proposed intelli- (2021) suggested the use of genetic algorithms-based
gent warehouse stocking using linear regression, random decision support systems to precisely set the lead time
and define the delivery route, thus reducing transporta- example, there is a need for a massive amount of clean
tion costs. data to train and test the models (Lee et al. 2018; Villalba-
Diez et al. 2019) and the need to apply complex hybrid
methods to deal with ambiguities and uncertainties typi-
2.5. Return
cal of real data (Khashei and Chahkoutahi 2021). How-
The return process includes ‘activities associated with ever, to the best of the authors’ knowledge, there is a
the reverse flow of goods. The processes associated with scarcity of contributions related to all issues that hinder
moving material from a customer back through the sup- AI implementation within real companies, which should
ply chain to address product defects, ordering, or manu- include not only technical but also financial and organ-
facturing, or to perform upkeep activities’ (APICS 2017). isational barriers. Therefore, it is useful to broaden this
For this reason, the application of AI to the return process analysis to the vast and consolidated body of literature
has been related to the concept of the circular economy on the implementation of Industry 4.0. Indeed, the bar-
(CE) and in particular, the CE component of reverse riers identified as critical by previous literature for the
logistics (Dhamija and Bag 2020; Li, Liu, and Li 2019; implementation of Industry 4.0 technologies can also be
Naz et al. 2022). Recent studies on reverse logistics have critical for AI, which is included among the plurality of
started focusing on the development of AI models to core technologies encompassed by Industry 4.0 (Bai et al.
optimise environmental factors, and machine learning 2020; Benitez, Ayala, and Frank 2020; Frank, Dalenogare,
algorithms have been proposed in the literature to sup- and Ayala 2019; Kamble et al. 2020; Osterrieder, Budde,
port CE (Noman et al. 2022). Indeed, reverse logistics and Friedli 2020). Thus, the analysis of barriers related to
are considered more complex to manage than forward the implementation of Industry 4.0 could be considered
logistics because many factors are unknown, such as the a good starting point for the empirical study of barriers
volumes, quality, and timing of product returns (Birkel to AI implementation. The main results are summarised
and Müller 2021). in Table 2.
According to a recent review published by Wilson, A recent study by Agostini and Filippini (2019) anal-
Paschen, and Pitt (2021), reverse logistics can be con- ysed the organisational and managerial challenges to
ceptualised in four main domains to which AI can con- Industry 4.0 adoption faced by Italian manufacturing
tribute: (i) network design, by evaluating potential third- firms through a survey research. They found that organ-
party logistics (3PL) providers and defining the optimal ising for Industry 4.0 requires training and continuous
number and location of collection points; (ii) collection, professional development of employees, lean production
by solving complex routing problems, estimating the practices and continuous improvement approaches in
number of articles returned during collection, and sup- production processes, the existence of a modern ICT
porting the gatekeeping function; (iii) warehousing, by infrastructure, and integration and collaboration among
forecasting return volumes and improving decision sup- supply chain partners increasing openness and adapt-
port for inventory management and supporting, sorting, ability. In addition, Pasi, Mahajan, and Rane (2020) con-
and inspection tasks; and (iv) processing, by selecting the ducted a literature review on IoT in supply chain manage-
most favourable processing option based on numerous ment and underlined that the challenges to smart supply
qualitative and quantitative factors. chain management are technology selection, data man-
The literature underlines the capability of AI for CE agement, cyber security, cultural resistance and struc-
and reverse logistics to provide more quantitative and tural problems. Also, Raj et al. (2020) studied barriers
robust solutions than other methods (Wilson, Paschen, inhibiting the adoption of Industry 4.0 based on a liter-
and Pitt 2021). In addition, AI in reverse logistics can ature review and expert input. In total, 15 barriers were
improve network design and collection by reducing identified, and the results highlighted how the impacts of
inventory carrying and holding costs without increasing these barriers are different in developed and developing
location costs and customer service, automating returns economies. The most important barrier for developing
sorting and making it faster, and increasing the environ- economies was the lack of standards, regulations, and
mental sustainability of processing options (Dutta et al. forms, and the low maturity level of the desired tech-
2021). nology for developed economies. However, the lack of a
digital strategy, alongside resource scarcity, has emerged
as the most prominent barrier in both developed and
2.6. Barriers to AI implementation in OSCM
developing economies.
Literature on barriers to the successful implementation The following studies deepened the analysis of these
of AI in OSCM is limited and related to technical issues barriers. Kumar, Vrat, and Shankar (2021) categorised,
related to the development of specific algorithms. For based on a literature analysis and expert input, 23
barriers to Industry 4.0 implementation. Stentoft et al. 2.7. Summary of the results
(2021) categorised ten barriers to Industry 4.0 imple-
From the analysis of the literature, various applications of
mentation and empirically analysed the implementation
AI techniques related to the five processes of the SCOR
of Industry 4.0 in small-medium enterprises (SMEs)
model and interesting benefits for companies have been
through survey and case research. Among them, the
underlined as a result of these applications. Table 1 shows
barriers most acknowledged by the companies anal-
a synthesis of the results obtained from the literature
ysed resulted in a lack of understanding of the strategic
review on AI applications for the different SCOR pro-
importance of Industry 4.0, lack of employee knowl-
cesses and the benefits achieved by companies owing to
edge about Industry 4.0, too much focus on operations
these AI applications. In particular, the literature review
at the expense of development, and lack of employee
shows that the main difference in the specific AI applica-
readiness. Moreover, Majumdar, Garg, and Jain (2021)
tions, identified in research studies, is related to the con-
focused on the study of barriers to Industry 4.0 imple-
text were the solutions proposed are applied. In the plan
mentation in SMEs in the textile and clothing industry.
process, AI solutions have been used for demand fore-
The study started with 22 barriers identified in the lit-
casting in the energy and food industry, and for inventory
erature and identified four driving barriers in the indus-
management the case of intermittent demand products
try and textile context: lack of awareness and commit-
and spare parts. In the source process, AI solutions have
ment of top management, lack of trained personnel,
been used for supplier selection in the electronic and gar-
lack of government support and policies and incen-
ment industry, and in the health sector, and for quality
tives for Industry 4.0, insufficient research and devel-
control of incoming orders in OEM and textile indus-
opment in Industry 4.0. Chauhan, Singh, and Luthra
try. In the make process, AI solutions have been used
(2021) categorised 20 barriers to Industry 4.0, and stud-
for scheduling, manufacturing, quality and maintenance
ied how they influence the linkage between digitalisa-
in manufacturing and production industry, and schedul-
tion and the firm’s performance in the emerging econ-
ing in the technology industry. In the deliver process,
omy context of India through survey research. Their
AI solutions have been used for stocking, picking, ship-
results indicate that Industry 4.0 adoption is dependent
ping and process inquiry and quote in the warehousing
on extrinsic barriers such as resistance by other stake-
and logistics industry, for process inquiry and quote in
holders, lack of standard architecture, lack of Internet
electronic industry and logistic optimisation in agricul-
access, contractual and privacy and security issues, and
ture industry. In the return process, AI solutions have
regulatory underdevelopment. In the context of man-
been used for network design, optimisation of collec-
ufacturing companies in India, Nimawat and Gidwani
tion activities, optimisation of warehousing and process-
(2021) explored the barriers relevant to Industry 4.0,
ing selection in logistics industry focusing on reverse
identifying 15 barriers from the literature, and anal-
logistics.
ysed their causal relationships using the DEMATEL tech-
The literature on barriers to AI implementation has
nique, finding that the implementation cost has the
not been covered by recent studies, and only a few tech-
maximum effect on all the other barriers. In the con-
nical barriers related to the development of specific ML
text of Italian manufacturing companies, Pozzi, Rossi,
algorithms have been identified. Thus, this study anal-
and Secchi (2021) analysed the factors affecting the
ysed the literature on barriers related to the implementa-
implementation of Industry 4.0 through case research.
tion of Industry 4.0, which includes AI among its tech-
Their results underlined the importance of continu-
nologies. Table 2 categorises and lists all the barriers
ous improvement/lean culture, explicit objective recog-
identified in the literature related to the implementation
nition, top management project leadership, multifunc-
of Industry 4.0.
tional project teams, clear and defined project plans, and
formal and specific training programmes for the suc-
cessful implementation of Industry 4.0. Finally, another 2.8. Research gaps, problem statement and
recent study in Italy was performed by Cugno, Castag- research objectives
noli, and Büchi (2021), who identified prominent barriers
to Industry 4.0 implementation through in-depth inter- This study is driven by the research gaps emerged from
views and multiple case studies. They categorised 11 bar- the review of the literature presented in the previous sec-
riers from the literature analysis, and their results high- tions, which, in line with recent studies on AI in OSCM
lighted four main barriers related to knowledge issues, (Cadden et al. 2022; Dhamija and Bag 2020; Pournader
economic-financial resources, cultural aspects, and sys- et al. 2021; Sharma et al. 2022; Tirkolaee et al. 2021), can
tem conditions. be synthetised as follows.
Table 1. Summary of the literature review on AI applications and benefits.

Generic AI application in
SCOR Processes Specific AI application in OSCM OSCM Benefits References
Plan • Energy industry: forecasting of electricity • Demand fore- • Growth of inventory Singh and Challa 2016; Khashei
consumption through cohort intelligence casting and turnover index and Chahkoutahi 2021;
& adaptive neural fuzzy inference system; planning • Reduction of the num- Tutun et al. 2022; Bala 2012;
electricity demand forecasting through • Inventory manage- ber of days out of stock Priyadarshi et al. 2019;
fuzzy hybrid intelligence-based seasonal ment • Optimal inventory con- Aktepe, Yanık, and Ersöz
models; electricity demand forecasting trol 2021; Lolli et al. 2019; Priore
through hybrid methods, such as adap- et al. 2019; Cantini et al.
tive neural-based fuzzy inference system 2022
(ANFIS) and fuzzy seasonal multilayer
perceptron (FSMLP)
• Food industry: demand forecasting at the
retail stage for selected vegetables through
long short-term memory (LSTM) and sup-
port vector regression (SVR).
• Intermittent demand products / spare parts
industry: demand and inventory manage-
ment of spare parts through artificial neural
network (ANN) and SVR.
Source • Electronic industry: green supplier selection • Supplier selection • Increase of the accuracy Sharma et al. 2022; Riahi et al.
through ANN with data envelopment anal- • Quality control of and decrease of the 2021; Kuo, Wang, and Tien
ysis (DEA) and a network analytic process incoming orders computation time in 2010; Gegovska, Köker, and
(ANP); different classifications Çakar 2020; Fashoto et al.
• Health sector: identify the best medical for supplier selection 2016; Bahadori et al. 2020;
equipment supplier using a combination of • High accuracy in Fallahpour et al. 2017; Luan
ANN and fuzzy VIKOR; defects identification et al. 2019; Yeh and Chuang
• Garment industry: supplier selection 2011; Hamdan and Jarndal
through a genetic-based intelligent 2017; Zhang and Cui 2019;
approach, gene expression programming Guo, Yuan, and Tian 2009;
(GEP); Islam, Amin, and Wardley
• Original Equipment Manufacturers: tempo- 2021; Mani et al. 2021;
ral difference learning approach to predict Zhao, Zhang, and Zhang
the acceptance and rejection paths of raw 2021; Unajan, Gerardo, and
materials; Medina 2019
• Textile industry: identify and classify differ-
ent types of defects through a convolutional
neural network (CNN)
Make • Manufacturing and production industry: Scheduling Reduction of costs Kim and Kim 2021; Wu et al.
forecasted machine breakdown using sim- Manufacturing Optimisations of 2020; Chan, Prakash, and
ulated annealing (SA) algorithm combined Quality resources usage (for Mishra 2013; Mohapatra,
with a fuzzy logic controller; dynamic dis- Maintenance instance, energy) Benyoucef, and Tiwari 2013;
patching of re-entrant production through Improved early fault Chen, Tai, and Hung 2012;
deep neural network (DNN) and Markov detection Lee et al. 2018; Lăzăroiu,
decision processes (MDP); allocations of the Improved afters-sales Kliestik, and Novak 2021;
machines based on the priority with arti- maintenance service Manimuthu et al. 2021;
ficial immune system (AIS) combined with Omwando et al. 2018;
a fuzzy logic controller (FLC); reduction in Wuest, Irgens, and Thoben
the production cycle time through artificial 2014; Carvalho et al. 2019;
immune system (AIS); define the best flow Kumar and Hati 2021; Luo
in a flexible job shop with decomposition- et al. 2019; Encapera, Gosavi,
integration genetic algorithm (DIGA); com- and Murray 2021; Kaparthi
bination of IoTs and sensors can bring and Bumblauskas 2020; Lee
for controlled and self-regulated produc- et al. 2018; Villalba-Diez
tion; real-time monitoring and optimisa- et al. 2019.
tion of resources and energy using k-means
algorithm; increase the accuracy and reduce
the cost of an industrial visual inspection
through deep neural network (DNN) soft
sensor; in predictive maintenance the use of
machine learning technics avoids unneces-
sary replacement of equipment, saves costs
and improves the safety, availability and
efficiency of processes.
• Tecnology industry: grouping products with
the similar customised using neural net-
work;
(continued)
Table 1. Continued.
Generic AI application in
SCOR Processes Specific AI application in OSCM OSCM Benefits References
Deliver • Warehousing and logistics industry: deter- • Stocking • Reduction of offer plan- Priore et al. 2019; Sharma
mine the best procurement policy reacting • Process inquiry and ning time and Singh 2021; Leung
to environmental changes by adopting an quote • Effective commu- et al. 2019; Kolomvatsos
inductive learning algorithm; predicting the • Shipping nication with the et al. 2016; Li and Xiao
demand and intelligent warehouse stock- • Picking customer 2021; Nowak, Hewitt, and
ing using linear regression, random for- • Logistics • Increase in offers accep- White 2012; Ellefsen et al.
est, and XGBoost; answer generic customer optimisation tance 2019; Hill, Jü, and Böse
enquiries and even automate direct sales • Increase in customer 2016; Loske and Klumpp
with natural language processing (NLP); a satisfaction 2021; Dosdoğru, Boru
B2B flexible pricing decision support system • Reduction of trans- İpek, and Göçken 2021;
applying fuzzy logic technique; optimisa- portation time and Shumanov and Johnson
tion of the splitting loads with tree models; costs 2021; Toorajipour et al. 2021;
• Electronic industry: automated concurrent Zdravković, Panetto, and
negotiations capable of performing simul- Weichhart 2022
taneous quotations through artificial bee
colony algorithm;
• Agriculture industry: optimise logistics and
product delivery methods using genetic
algorithms applied through data mining.
Return • Logistics industry: optimise the location • Network design • Reduction of inventory Li, Liu, and Li 2019; Dhamija
selection in the remanufacturing logistics • Optimisation of col- carrying and holding and Bag 2020; Naz et al.
network through ant colony optimisation lection activities costs 2022; Noman et al. 2022;
(MACO) algorithm; support reverse logistics • Optimisation of • Reduction of location Birkel and Müller 2021;
in circular economy with fuzzy logic and warehousing costs Wilson, Paschen, and Pitt
decision support system; ML technique in • Processing • Improvement of 2021; Dutta et al. 2021
the reverse logistics can contribute to eval- selection customer service
uating potential third-party logistics, solv- • Automation and speed
ing complex routing problems, forecast- up of returns sorting
ing return quantities, selecting the most • Increased environmen-
appealing processing option; improve net- tal sustainability
work design and collection using genetic
algorithm.
There is a scarcity of studies in the literature address- other studies considered ML methods in supplier selec-
ing the way how organisations should implement AI tion along with the problem of order allocation holisti-
techniques for supporting OSCM processes. Few mod- cally. Furthermore, the literature review on AI, ML, and
els have been proposed to guide companies that want to DL in smart logistics by Woschank, Rauch, and Zsifkovits
implement and use AI to support OSCM processes. (2020) underlines that most studies in the field are either
There is a lack of exploratory empirical in-depth stud- conceptual, in a very early testing phase, or laboratory
ies that proposed AI techniques to support OSCM pro- experiments.
cesses, such as multiple case studies research. Most stud- In line with the abovementioned research gaps, we can
ies addressed this topic through not empirical methods, conclude that the problem addressed by this research is
performing bibliometric or systematic literature reviews, the lack of empirical studies in the literature that analyse
or testing new AI applications with simulated data. in-depth industry case experience with AI. The conse-
There is a dearth of evidence from mature industrial quence is that it is difficult for practitioners to find guide-
applications and the generalisability of the AI techniques lines to implement AI within different industrial reali-
proposed is not assessed today. Most studies that applied ties and understand the applicability of AI techniques to
AI solutions to real cases addressed single experimental OSCM. Therefore, this study conducts empirical quali-
cases, focusing on a single application to a single sector. tative research through multiple case studies, aiming to
In addition, this study based its analysis also consid- achieve the following objectives: (i) Understand how AI
ering the results obtained from recent literature review applications support OSCM; (ii) Comprehend what are
studies on AI applied to different OSCM processes, which the benefits achieved by companies when using AI to sup-
highlighted and confirmed similar issues. For example, port OSCM; (iii) Acquire what are the barriers tackled by
Abbas Naqvi and Amin (2021), after reviewing 92 papers companies when implementing AI to support OSCM.
related to supplier selection, mentions the possibility that
advanced deep learning (DL) and ML techniques can
3. Methodology
be applied to forecast the parameters of the optimisa-
tion models only in future research. Islam, Amin, and This study addresses the research gaps and answers the
Wardley (2021) highlighted that before their study, no research questions by combining the findings of the
Table 2. Summary of the literature review on barriers to Industry 4.0 implementation.

Category ID Barriers Reference
Financial barriers 1 High investment needed to implement Industry 4.0 Raj et al. 2020; Kumar, Vrat, and Shankar 2021; Stentoft
initiatives, insufficient financial resources within the et al. 2021; Majumdar, Garg, and Jain 2021; Chauhan,
company and scarcity of external financing Singh, and Luthra 2021; Cugno, Castagnoli, and Büchi
2021; Nimawat and Gidwani 2021
2 Lack of clarity regarding the economic benefit and Raj et al. 2020; Kumar, Vrat, and Shankar 2021; Majumdar,
low experience in budgeting, cost analysis, resource Garg, and Jain 2021; Chauhan, Singh, and Luthra 2021;
allocation, etc. for Industry 4.0 Cugno, Castagnoli, and Büchi 2021; Nimawat and
Gidwani 2021
3 Lack of risk management tools to decide the amount of Majumdar, Garg, and Jain 2021
investments in Industry 4.0
Organisational barriers 4 Enhanced skills required for employees, shortage of Raj et al. 2020; Agostini and Filippini 2019; Kumar, Vrat,
workforce with adequate skillset in the labour market and Shankar 2021; Stentoft et al. 2021; Majumdar, Garg,
and lack of internal training and Jain 2021; Chauhan, Singh, and Luthra 2021; Cugno,
Castagnoli, and Büchi 2021; Nimawat and Gidwani 2021;
Pozzi, Rossi, and Secchi 2021
5 Lack of internal digital culture and training to embrace Raj et al. 2020; Agostini and Filippini 2019; Kumar, Vrat,
technological advancements and Shankar 2021; Stentoft et al. 2021; Majumdar, Garg,
and Jain 2021; Chauhan, Singh, and Luthra 2021; Cugno,
Castagnoli, and Büchi 2021; Nimawat and Gidwani 2021
6 Lack of employees’ readiness and high resistance to Raj et al. 2020; Kumar, Vrat, and Shankar 2021; Stentoft
change et al. 2021; Chauhan, Singh, and Luthra 2021; Cugno,
Castagnoli, and Büchi 2021; Pasi, Mahajan, and Rane
(2020)
7 Lack of knowledge management systems and low Raj et al. 2020; Agostini and Filippini 2019; Majumdar, Garg,
integration and collaboration among organisation and Jain 2021; Chauhan, Singh, and Luthra 2021; Pozzi,
departments and supply chain partners Rossi, and Secchi 2021; Pasi, Mahajan, and Rane (2020)
Strategical barriers 8 Difficult process changes and ineffective change Raj et al. 2020; Majumdar, Garg, and Jain 2021; Chauhan,
management Singh, and Luthra 2021; Nimawat and Gidwani 2021;
Pozzi, Rossi, and Secchi 2021
9 Lack of strategical roadmap for Industry 4.0 and weak Raj et al. 2020; Kumar, Vrat, and Shankar 2021; Majumdar,
methodical approach for implementation Garg, and Jain 2021; Chauhan, Singh, and Luthra 2021;
Nimawat and Gidwani 2021; Pozzi, Rossi, and Secchi
2021
10 Lack of top management commitment Kumar, Vrat, and Shankar 2021; Stentoft et al. 2021;
Chauhan, Singh, and Luthra 2021; Nimawat and Gidwani
2021; Pozzi, Rossi, and Secchi 2021
11 Lack of lean production practices and continuous Agostini and Filippini 2019; Pozzi, Rossi, and Secchi 2021
improvement approaches
12 Lack of stakeholders’ involvement and engagement Kumar, Vrat, and Shankar 2021; Chauhan, Singh, and Luthra
2021
13 Vagueness in understanding Industry 4.0 and its strategic Kumar, Vrat, and Shankar 2021; Stentoft et al. 2021;
importance Majumdar, Garg, and Jain 2021; Cugno, Castagnoli, and
Büchi 2021; Nimawat and Gidwani 2021
14 Lack of collaboration between academic institutions and Stentoft et al. 2021; Kumar, Vrat, and Shankar 2021; Cugno,
industry for innovation Castagnoli, and Büchi 2021
15 Too much focus on operational aspects and poor R&D on Stentoft et al. 2021; Majumdar, Garg, and Jain 2021;
Industry 4.0 adoption Nimawat and Gidwani 2021
Technological barriers 16 Lack of basic physical and digital IT infrastructure for Raj et al. 2020; Agostini and Filippini 2019; Kumar, Vrat, and
data driven services and poor provision of broadband Shankar 2021; Majumdar, Garg, and Jain 2021; Chauhan,
infrastructure and connectivity Singh, and Luthra 2021; Cugno, Castagnoli, and Büchi
(2021); Nimawat and Gidwani 2021; Pasi, Mahajan, and
Rane (2020)
17 Integration issues like compatibility, scalability and inter- Kumar, Vrat, and Shankar 2021; Majumdar, Garg, and Jain
operability between different machines, equipment, 2021; Cugno, Castagnoli, and Büchi 2021; Nimawat and
technologies and network systems Gidwani 2021; Pasi, Mahajan, and Rane (2020)
18 Low maturity level of the desired technology Raj et al. 2020
19 Resistance to share data due to lack of information security Raj et al. 2020; Kumar, Vrat, and Shankar 2021; Stentoft
and privacy protection et al. 2021; Majumdar, Garg, and Jain 2021; Chauhan,
Singh, and Luthra 2021; Nimawat and Gidwani 2021;
Pasi, Mahajan, and Rane (2020)
20 Challenges in ensuring data quality Raj et al. 2020; Chauhan, Singh, and Luthra 2021; Pasi,
Mahajan, and Rane (2020)
Legal barriers 21 Lack of legal regulations to manage cybercrime and data Kumar, Vrat, and Shankar 2021; Majumdar, Garg, and
theft Jain 2021; Chauhan, Singh, and Luthra 2021; Cugno,
Castagnoli, and Büchi (2021); Nimawat and Gidwani
2021; Pasi, Mahajan, and Rane (2020)
22 Lack of standards and government regulations embracing Raj et al. 2020; Kumar, Vrat, and Shankar 2021; Stentoft
Industry 4.0 regarding labour and employment and data et al. 2021; Majumdar, Garg, and Jain 2021; Chauhan,
ownership and copyright Singh, and Luthra 2021; Cugno, Castagnoli, and Büchi
(2021); Nimawat and Gidwani 2021
Figure 1. Summary of the research steps.
literature review with direct empirical observations 2015). A further value of case studies lies in the possibility
through a case study methodology (see Figure 1). This of checking the validity of responses due to the nature of
qualitative research methodology, known to be ‘oriented personal communication and interviewees, who are most
towards exploration, discovery and inductive logic’ (Pat- certainly experts on the subject (Blome and Schoenherr
ton 2002), proves to be ideal for developing or broaden- 2011).
ing theories and providing empirical insights when the According to recent studies on case study research
knowledge on a topic is scarce or few studies have been (Voss, Johnson, and Godsell 2016; Robert K Yin 2018),
published on it (Jabbour et al. 2015). For AI, a topic this study rigorously applies the following methodolog-
that is certainly widely discussed but with still limited ical steps: (i) case study design; (ii) case selection; (iii)
applications and only scarce empirical research in OSCM data collection and analysis. To ensure the reliability of
(Dubey et al. 2020; Helo and Hao 2021; Pournader et al. the research these steps are presented in detail in the
2021), this methodology seems to adapt perfectly. following subsections.
Case studies provide practical results that will not only
help understand the topic, but also provide elements to
3.1. Case study design
guide its practical application (Eisenhardt 1989; Lisa M.
Ellram 1996; R K Yin 2009). Moreover, case studies can The methodology applied in this study involves multi-
provide detailed information that can go unnoticed in ple case studies. This methodology allows the authors
aggregate and top-down quantitative analyses (Carrillo- to increase the external validity of the results owing
Hermosilla, del Río, and Könnölä 2010; Jabbour et al. to the possibility of collecting data from multiple cases
(Robert K Yin 2018). To limit variation and control for for the selection of the cases: following the literal repli-
the research domain (Voss, Johnson, and Godsell 2016), cation, AI applications were selected based on the expec-
this study addressed a population of companies operat- tation that they tackle similar barriers and have similar
ing in Italy. According to a recent report provided by the benefits, since they are all applied by companies operating
Italian Government (2022–2024), AI has vast potential, in Italy, within the same economic and cultural context;
which is still not fully exploited. This report shows that following the theoretical replication, AI applications were
there are a considerable number of players in the Italian selected based on the expectation that they face different
market offering AI products, which are general in nature contextual variables while being implemented, since they
or for specific applications, and 53% of medium-large are all applied by companies that are heterogeneous in
Italian companies declare that they have started at least terms of sector, number of employees, and turnover. The
one AI project, mostly belonging to the manufacturing information for selection was found on the companies’
sector. Therefore, companies operating in this country websites, universities, and consultancy reports. More-
and successfully applying AI to support OSCM are par- over, the authors had experience with these companies
ticularly interesting to study, as they tackle barriers when because of their previous research projects. In total, six
implementing AI, but at the same time, can overcome companies and 17 AI applications were selected. Table 3
them while many other companies are struggling. Thus, provides an overview of the case study.
this study aims to analyse the AI solutions that compa-
nies operating in Italy are applying to support OSCM and
3.3. Data collection and analysis
understand the insights that can be gained on the benefits
and barriers of AI in this field. It is possible to find more Data collection was carried out through semi-structured
than one AI application within the same company; thus, face-to-face interviews to maintain a logical order while
the focus of this study and the unit of analysis is the AI ensuring spontaneous deviation and open comments.
application. The semi-structured interview, despite being based on
subject directions found prior to the interview, allows the
researcher to go in-depth for a discovery (Magaldi and
3.2. Case selection
Berler 2020). It offers both flexibility and the possibility
The complete list of companies operating in the Ital- to go further into topics while maintaining the focus of
ian manufacturing and service industries was extracted the study (Ruslin et al. 2022).
from the database ‘AIDA’ (https://aida.bvdinfo.com/), In this research, the participants were experts in
and only big companies (according to EU recommenda- AI (e.g. maintenance specialists, data science directors,
tion 2003/361) were selected because, from preliminary and head of production process engineering) directly
analyses, investments in AI require economic resources involved in or responsible for AI projects in their com-
and skills that are more readily available in these types pany and fully informed of the ongoing situation and
of companies. The unit of analysis of the study was the problems. Each interview lasted approximately 120–180
IA project because a company can apply several differ- minutes. In each company, there were at least two inter-
ent solutions to overcome specific challenges. Within this viewees (see Table 3) to improve the validity (R K Yin
predetermined list, the selection of cases accurately fol- 2009) and reliability of the collected data (Voss, Tsikrikt-
lowed theoretical rationales, based on the logic of repli- sis, and Frohlich 2002). All interviews were recorded and
cation (Miles and Huberman 1994). transcribed verbatim.
A total of six Italian companies were selected. These The purpose of this study and the literature review
were considered an adequate number according to the linking AI and the OSCM processes formalised in the
logic of applied replication. In fact, the companies pro- SCOR model reported in section 2 led the authors
vided 17 cases and an appreciable amount of literal and to define an interview protocol consisting of questions
theoretical replications that will be presented in the fol- regarding: (i) general information about the company
lowing sections, such as different and contrasting chal- and the interviewee, (ii) details on benefits achieved when
lenges to be faced, focus on different processes of the using AI to support OSCM, and (iii) details on barriers
SCOR model, as well as several different AI tools and tackled when implementing AI to support OSCM. The
techniques adopted. These sample characteristics rein- interview protocol is reported in Appendix 1.
force and support the findings of the study (Robert K Yin Furthermore, archival data, including annual reports,
2018). websites, and documents on AI projects, were used to tri-
A theoretical sampling based on replication logic angulate empirical evidence and ensure multiple sources
(Eisenhardt 1989; Miles and Huberman 1994) was used of evidence (Hays 2004; Stuart et al. 2002).
Table 3. Case study overview.

Turnover 2021 Number of Number of Respondents’ Case study: AI
Company ID [million e] employees 2021 interviewee/s corporate role Sector application
A 2,198 4,654 2 Production Development Electronic components, A1 = Visual Inspection
Specialist automation and for in process quality
Maintenance Specialist robotics control
A2 = Visual Inspection for
product quality control
A3 = Intelligent Cobot
A4 = Predictive
Maintenance
B 768 2,100 2 Data science Director Manufacturer of pipes B1 = Prediction of
and related services industrial cost
for the world’s energy B2 = Anomaly detection
industry system
B3 = Virtual sensing
system for scrap
recovery
C 200 800 2 Maintenance Engineer Light firearms for C1 = Classification of
Responsible for Engi- hunting, sporting and downtimes
neering of Production personal defence
Processes C2 = Predictive
Maintenance
C3 = Visual Inspection for
product quality control
D 1,647 10,772 4 Program Manager Manufacturer of electrical D1 = Anomaly detection
and semiconductor system
components D2 = Intelligent batch
dispatching system
D3 = Virtual inspection
for in process quality
control
D4 = Correlation analysis
on electrical test results
E 71 536 2 HR Manager Textile E1 = Optimisation of raw
Supply Chain Manager material replenishment
F 4,171 1,130 2 Operations Process Oil and Gas F1 = Analysis and
Manager planning of the
‘rebound’
F2 = Intelligent interactive
voice responder
3.4. Validity and reliability 1989; Miles and Huberman 1994), detailed in section 3.1,
and highlighted the similarities and differences in the
This study aims to improve the validity and reliability of
cases analysed.
the analysis by adopting several strategies suggested by
Eisenhardt (1989) and Yin (2009).
To ensure construct validity, we triangulated empir- 4. Results
ical evidence from multiple sources, as described in
section 3.3. All interviews were recorded. For each case, 4.1. AI applications support supply chain and
a database was created containing verbatim transcripts of operations management processes
the interviews, archive data and field notes. Table 4 presents a summary of the AI applications iden-
To ensure reliability, a protocol was followed during tified by the companies analysed. The applications found
the fieldwork (see Appendix 1). Two authors manually within the cases are related to the plan, make, and deliver
conducted the coding and analysis of the interview and processes. Details are presented in the following sections.
secondary data for each case; the few divergences in
the initial codification were extensively discussed until
an agreed final decision was reached, as suggested by 4.1.1. Plan processes
Duriau, Reger, and Pfarrer (2007). Concerning demand planning and forecasting, Company
Internal validity was ensured by iteratively compar- F used to face a particularly critical situation due to a
ing the results of the analysis with the literature and the concentration of calls during customer billing periods.
proposed model until theoretical saturation was reached. In those periods, there was an excessive lengthening of
To enhance the external validity, and thus the general- the waiting time before being able to talk to an opera-
isability of the results, we adopted a theoretical sampling tor, which generated a lot of complaints from customers.
method based on the logic of replication (Eisenhardt Therefore, they launched a project (F1) focused on the
Table 4. Results on AI applications in OSCM.

SCOR Company [X] and
Processes AI application in OSCM Details on AI applications AI solution [#]
Plan Demand planning Analysis of historical demand data to forecast future customer demand to optimise the planning F1
and forecasting of the resources, plan marketing actions and balance workloads.
Inventory Analysis of raw material price trends, social media trends, and company’s offer, to optimise raw E1
management material replenishment and achieve a good inventory management.
Make Scheduling Analysis of conditions in the production department and resources’ availability to rout batches in D2
real-time to the various machines.
Manufacturing Analysis of operators’ gesture through visual control combined with cobot capabilities, to perform A3
repetitive tasks, inspection, and quality control activities.
Analysis of variables related to systems behaviour to keep the production process constantly under B2, D1
control, set process limits and solve production problems detected in real time.
Analysis of chemical elements within the scraps to optimise the recycling of scrap products, B3
minimising the scraps purchasing cost and finding the best mix of scraps that complies the
quality needed for the production system.
Quality Analysis of process images or videos for in process quality control, to ensure process quality and A1, D3
automation, substitute the repetitive and alienating human control, and support root-cause
analysis and decision-making.
Analysis of product images or videos for product quality control, to measure the output of A2, C3
production, compare it to the production target and previous production.
Analysis of product functioning to analyse the correlation between the physical measurements D4
and the functioning of devices that fail tests and recognise configurations that can lead to test
failures.
Maintenance Analysis of machine functioning through sensors’ data to identify the degree of ‘health’ and the A4, C2
degradation index of each individual machine (or its components), estimate the remaining
useful life of the machines (or its components) and predict failures.
Analysis of machine downtimes collected manually and through sensors to classify potential C1
causes and support maintenance activities.
Deliver Process inquiry and Analysis of historical data to identify relationships between industrial costs and product B1
quote characteristics to predict future industrial costs of new build-to-order products and generate
reliable quotations.
Analysis of speeches to improve automatic customer care and quickly and intuitively understand F2
how to approach customer needs and problems.
analysis and planning of the rebound, which is the peak trends related to colours or shapes); and (iii) customer
of calls that the call centre receives from customers fol- samples (e.g. group of silk product samples offered to
lowing the issuing of electricity and gas bills. The project the buyers for presenting newly developed styles). This
leverages an ML algorithm that examines the ‘rebound’ application has been used to balance stock-holding costs
history to forecast future customer calls, optimise the and customer service levels in a sector affected by high
planning of the resources needed by the call centre to demand variability.
cover the possible demand curve, and guarantee a high
customer service level. Additionally, this application has 4.1.2. Make processes
been used to plan marketing actions and scheduling As far as the scheduling process is concerned, for Com-
when it is more effective in sending emails or messages pany D the optimal planning of the batch launch becomes
about latest offers to avoid proposing commercial deals a key issue to guarantee adequate results in terms of effi-
that could generate an overload of lines and operators. ciency and effectiveness. In the past, various solutions
Concerning inventory management, Company E, like based on classical mathematical modelling tools have
many other companies in the fashion industry, faces a been developed and tested; however, these applications
very specific problem. Fabric replenishment orders have have shown major limitations as queues were forming
very long lead times, and a lack of availability imme- in the departments, slowing down the production flow.
diately turns into potential order losses. Conversely, an The company developed a system for intelligent batch
excess of stock has a big impact on working capital dispatching (D2) based on the reinforcement learning
and high obsolescence costs. They developed a project algorithm. This allows batches to be routed in real time
(E1) for the optimisation of raw material replenish- to various machines based on the conditions in the pro-
ment and consequent good management of raw material duction department and their availability. After several
inventories. This is an ML algorithm that suggests how simulation trials, the system identifies different scenarios,
to place raw material orders based on the analysis of data including pessimistic scenarios, for example, a machine
on events that can generate a purchase order: (i) histor- being down for a long time. This process is continuous,
ical prices of raw materials (e.g. the trend of silk price); and the system defines and learns how to react to different
(ii) information taken from different social media (e.g. situations.
Regarding manufacturing concerns, Company A had a The introduction of an anomaly detection system based
critical issue with quality control activities in production. on ML has made it possible to analyse typical curves,
Since these controls were carried out in a robot work area, learn from them, compare the parameters detected, and
production had to be stopped. Furthermore, controls provide real-time engineering with an answer to the
were visual and depending on who was doing the con- process.
trol, different outcomes could be obtained. They imple- Company A introduced two visual inspection meth-
mented a project (A3) concerning Intelligent Cobot, that ods to improve the quality of the production process.
is, a collaborative robot combined with AI, which can The first visual inspection tool (A1) operates on a robotic
recognise operators’ gestures through visual control. The island that manages the assembly and screwing of a com-
cobot is used for repetitive tasks with low loads as well as ponent. Before the component is closed and screwed on,
for inspection and quality control activities. Company B it must be greased by the robot, which autonomously
developed two different projects (B2 and B3) in the man- takes the piece and places it on top of the greasing tools. In
ufacturing environment. When the furnace is restarted this case, the AI application analyses images taken from
after a routine maintenance operation, the process lead- the production process and assesses the amount of grease,
ing to the ideal operating temperature is particularly crit- and if it is placed correctly and in the correct positions.
ical and could have lethal effects on people working at This allows the robot to work autonomously without
the site. Therefore, they designed an anomaly detection human intervention, preventing an operator from per-
system (B2) based on a neural network that is trained sonally checking each piece and substituting repetitive
using data collected by the system’s temperature sensors. and alienating human control. The second visual inspec-
This allows the estimation of the ideal operating tem- tion tool (A2) was applied to the switches’ quality gate
perature of the melting furnace and compares it with its to measure the switches’ insertion distance and torque.
actual value, thus maintaining the entire process under To assist this application, the company used a series of
control. When a significant deviation is recorded between photographs, depicting the insertion distance; the pho-
the data collected by the sensors and the ideal tempera- tographs were analysed and used as input data using an
ture predicted by the AI system, an early warning alarm is ML tool that compares them to previous images and the
triggered, alerting operators to stop production, as there production target, and defines the result of the quality
may be problems with product features or, in extreme test. In addition, Company C launched a visual quality
cases, a risk of explosion. The second AI application control project (C3) because they needed to carry out an
(B3) was developed because they are not able to accu- objective quality control to determine an adequate value
rately determine the composition of the scrap used. The for the weapon. An overestimated valuation can lead to
presence of low-quality scrap is indispensable for lower- damage the brand reputation, while an underestimated
ing the average cost of production; on the other hand, valuation can lead to a substantial loss of turnover. They
such scrap contains a high amount of polluting materials. installed a machine equipped with cameras within the
The application is a virtual sensing system based on ML over-and-under rifle line. This solution is based on deep
and estimates the content of chemical elements within learning algorithms; it can scan and recognise the quality
the scraps. Based on these input data, the best combi- of the wooden stocks of rifles by comparing them with
nation of scraps of different quality (e.g. high-quality models from a historical archive. Company D has devel-
scrap, with very few pollutants, is represented by railways, oped two quality control projects. The first AI project
while extremely low-quality scrap is represented by tuna (D3) applied image recognition to optical scans for qual-
cans, peeled tomatoes cans, etc., because they contain a ity control and defect testing during the production pro-
lot of tin). The AI application is used to optimise the cess of the wafer, a thin slice of semiconductor material,
recycling of scrap products, minimising the scrap pur- with defect positions. Subsequent analysis conducted by
chasing cost and finding the best ‘recipe’ that complies process engineering is required to understand how to
with the quality needed for the production system. In treat the wafers, as not all defects necessarily lead to prod-
addition, Company B developed a project related to an uct rejection. An algorithm associated with image recog-
anomaly detection system (B1) that analysed the data nition supports root-cause analysis and aids in decision
detected and collected by the machine sensors for tem- making. The second quality control project (D4) involved
perature, pressure, gas flow, and electrical current used electrical testing, which was the ultimate step in the pro-
to check if the process is operating correctly. By analysing duction process. The system is based on supervised and
the standard curves of these parameters, it is possible to unsupervised ML algorithms to analyse the correlation
set limits that, if violated, categorise the process as a ‘crit- between the results of electrical tests and physical mea-
ical process’. These limits are extraordinarily complex to surements of devices that fail these tests. This would
set, because various devices may have different responses. make it possible to recognise configurations that can lead
to test failures before they reach the final stage and avoid three to four compulsory steps by typing the keys on
work on failing devices. the telephone several times before being able to speak to
An area of particular interest in AI applications is the operator. The new tool capable of identifying with
maintenance, mainly predictive maintenance, which is an accuracy of 85% the customer on call by matching
applied to automated processes. Company A developed a the phone number with the one recorded in the com-
predictive model (A4) capable of estimating the remain- pany database and using a series of disambiguation ques-
ing useful life of the machines (or their components) tions such as customer number, first and last name, and
and predicting failures to achieve benefits in terms of address. The special feature of this application allows the
costs and time. One of the components analysed was customer to provide his/her personal data by voice, like a
the electric motor, which provided a large amount of normal conversation with an operator, instead of dialling
data, including temperature, torque absorbed, current, a series of numbers and information, as done in the past.
vibration, number of cycles, and maximum speed. Com- The AI voice responder understands the caller’s basic
pany C realised two algorithms based on ML techniques needs and decides whether to forward the call immedi-
that were applied to several machines in the plant. The ately to the call centre or try to resolve the problem on its
first algorithm (C1) determines the causes of down- own and, only if this is not possible, to continue the call
time and classifies them by type. Data from the MES with an operator.
and ERP were used as inputs, as well as data provided
by the maintenance managers, including the output of
4.2. Benefits achieved by companies when using AI
another algorithm indicating whether the machines were
to support supply chain and operations
operational or stopped at a given time and how long
management
they had been in one of the two states. This algorithm
achieved reliability and accuracy of approximately 98%. The companies analysed highlighted the achievement of
The second application is a predictive maintenance tool several benefits thanks to the use of AI to support OSCM.
(C2), which has been adopted for more than 60% of Company A declared that the A1 project made it pos-
the company’s machinery. This solution is based on a sible to improve the safety and reliability of processes
system called ‘fingerprint’ that allows the creation of a and reduce lead times owing to the automation of human
digital fingerprint capable of indicating the degree of controls, whereas A2 established standard conditions to
‘health’ and the degradation index of each individual impartially verify whether a product corresponds to the
machine, as well as the performance related to it. This reference model, generating a saving of 1.5 minutes per
makes it possible to intervene in the entire equipment, unit during quality control and increasing quality control.
or only on a part of it, if it does not achieve the expected Applying AI to cobot (A3), instead, opened the possi-
performance. bility of interacting with the robot via gestures, allow-
ing the ‘gaming’ effect. Through this new functionality,
4.1.3. Deliver processes the company has achieved an increase in overall equip-
As far as the delivery process is concerned, the compa- ment effectiveness (OEE) and improved safety for oper-
nies analysed focused on their AI applications to support ators working alongside the robot. Finally, the predictive
process enquiry and quote. Company B implemented an maintenance project (A4) made it possible to reduce the
algorithm (B1) that predicted the cost of build-to-order lead time and maintenance costs, as well as improve
products. In this case, the company does not know the the OEE.
cost of producing a product that it has never done before, Company B highlighted that project B1 allowed the
and it is difficult to collect data to make these estimates. firm to increase the reliability of the cost estimation,
Furthermore, analysts starting with the same customer which reduced the response time of the quoting system
request can generate different quotations. Finally, ana- from days to one-two seconds. In addition, human errors
lysts cannot estimate the consistency of a proposal. ML were significantly reduced, and the reliability of the quo-
algorithms are objective and repeatable and can estimate tation increased, with a consequent positive effect on the
uncertainty. To implement the solution, the firm col- service level. The second project (B2) increases safety
lected a very large database of the actual costs of items thanks to real-time monitoring of machines and simul-
already produced and trained the model to identify the taneously increases process reliability and OEE owing
relationships behind the data. to the instant and quick intervention in the problems.
Additionally, to improve customer care, company F Finally, the third project (B3) optimises the mix of scraps
has developed an intelligent interactive voice respon- and reduces the purchasing costs of materials while
der (F2). In the past, the company used a traditional ensuring high environmental sustainability by recycling
answering system, and the customer was forced to make scrap.
Table 5. Results on benefits of AI applications in OSCM.

Plan Make Deliver
Demand
planning Process
and Inventory Inquiry and
forecasting Management Scheduling Manufacturing Quality Maintenance quote
Performance Performance
Category metrics F1 E1 D2 A3 B2 D1 B3 A1 D3 A2 C3 D4 A4 C2 C1 B1 F2 Total
Cost Reduction of X X X X X X X X 15
costs
OEE X X X X X X
improvement
Stock rotation X
improvement
Speed Reduction of lead X X X X X X X 9
times
Process reliability X X
improvement
Dependability Service level X X X X 4
improvement
Quality Increased X X X 3
product
quality
Safety Safety X X X 3
improvement
Sustainability Environmental X 1
sustainability
improvement
Total 5 1 8 9 8 4
For Company C, the application of ML to mainte- delivery times. This favours operational benefits with low
nance (C1 and C2) made it possible to organise produc- purchasing, high rotation turnover for raw materials, and
tion with a view to necessary or planned maintenance an increased service level. From a quantitative perspec-
work and reduced the average time to repair a fault by tive, the application of the ML algorithm resulted in a
approximately 30 minutes. They also generated a 4% 30% reduction in stock.
gain in machine availability, thereby improving the OEE. Finally, Company F presented the F1 project as an
Moreover, quality control through the AI application application that allowed the internal resources of the call
C3 increased product quality by reducing the reliability centre to be saturated, keeping costs low, and improved
of controls and the ‘false positive’ results in the quality the management of peaks, reducing unmet demand, and
check. increasing service level. Owing to the use of the AI
Company D stated that project D1 enhanced the root- algorithm, it is possible to reduce the abandonment rate,
cause analysis and supported the improvement of the minimise disruption, and reduce operating costs. The
finished product quality, an essential performance in the advantage of the intelligent interactive voice responder
semiconductor industry. The system for the intelligent (F2) is that it enriches communication and humanises
batch (D2) increased the number of batches leaving the an interaction with the customer that was previously
production line, creating a more continuous and constant cold and aseptic. This application improves the customer
flow. The benefit of the image recognition system (D3) experience by reducing waiting time when switching to
was cost reduction due to the reduction in scrap rate, as an operator and reducing operating costs, as a call han-
not all defective wafers necessarily have to be eliminated, dled by the call centre costs much more than a voice
and some of them can be recovered. The fourth project responder call.
(D4) allows a considerable reduction in costs, as well as Table 5 presents a summary of the main results
more time and margin of action to implement possible obtained from the cases. Since the benefits have been
safeguards and countermeasures. measured by companies through performance metrics
Company E underlined that project E1 made more that were improved thanks to AI projects, the results
accurate order forecasting possible and allowed for the were categorised, as suggested by Slack, Chambers,
optimal investment of available capital based on price and Johnston (2010), into the operational performance
trends and transportation costs, being able to prioritise dimensions. In particular, cost resulted as the dimension
certain materials and fabrics that are difficult to find that took the most advantage from AI projects, followed
and for which stock availability directly affects customer by speed, dependability, and quality. Additionally, a small
Table 6. Results on barriers to AI implementation in OSCM. information, as well as the huge investment needed
Category ID Barriers Companies to conduct the data capture process without a clear
Financial barriers 1 High investment needed to A, E, F return on investment (ROI). Furthermore, with the
implement AI initiatives, implementation of the AI cobot (A3), Company A
insufficient financial
resources within the
experienced some preliminary mistrust from operators
company and scarcity of towards a new technology that had never been used
external financing before.
2 Lack of clarity regarding A, C, F
the economic benefit Company B has faced difficulties in creating a con-
and low experience in sistent database. It is based on the consistency of the
budgeting, cost analysis,
resource allocation, etc. database that the effectiveness of an AI system derives:
Organisational 4 Enhanced skills required A, B, C, D, F the higher the quality of the data collected, the more the
barriers for employees, shortage
of workforce with developed model will be able to return truly useful out-
adequate skillset in the put in terms of predictions or suggestions. Another point
labour market and lack
of internal training
of focus is the reliability of the system. Often, the AI sys-
5 Lack of internal digital A tem is erroneously considered to always be reliable, and
culture and training to as a result, the suggested solution is accepted uncriti-
embrace technological
advancements cally. Instead, it must be considered that, even though
6 Lack of employees’ A, F the AI model has undergone an adequate learning pro-
readiness and high
resistance to change cess, there is still a probability that it may provide answers
7 Lack of knowledge C, D that are not necessarily correct. Another critical aspect
management systems
and low integration that was emphasised concerns the availability of specific
and collaboration knowledge and skills, both concerning the internal pro-
among organisation
departments and supply
duction processes being addressed and development of
chain partners AI models.
Strategical barriers 8 Difficult process changes D Company C highlighted the lack of cross-competen-
and ineffective change
management cies to build an interfunctional team dealing with such
Technological 17 Integration issues like E projects. To overcome this problem, all AI projects have
barriers compatibility, scalability
and interoperability been implemented by combining internal and external
between different expertise from consulting firms and universities. Inter-
machines, equipment,
technologies, and
viewees highlighted a second barrier in assessing the
network systems possible benefits derived from the adoption of AI tools
20 Challenges in ensuring A, B, D, E, F compared to the implementation costs incurred without
data quality
a clear ROI. Another barrier encountered by the company
is the lack of replicability due to the bottom-up AI project
number of cases underlined safety and sustainability as approach.
further benefits. Company D focused on three main issues in the devel-
opment of AI projects. The first is related to data in terms
of the complexity of the data collection, preparation,
4.3. Barriers tackled by companies when
and storage processes. The company had to develop an
implementing AI to support supply chain and
effective infrastructure to support the collection, prepa-
operations management
ration, and cleaning of data from various sources. The
The results of this study underlined nine main bar- data management process requires a great deal of stan-
riers tackled by companies when implementing AI dardisation because the production equipment is pro-
projects, belonging to the financial, organisational, strate- vided by different suppliers and follows different stan-
gic, and technological categories. Table 6 summarises the dards, which increases the data management. The other
obtained results, which are detailed below. two barriers concern organisational hurdles. The first is
The main barriers faced by Company A lie in the establishment of a change-oriented corporate mind-
the high complexity of data collection caused by the set and culture, as not all internal resources have the
variability of the products offered, both in terms of skills necessary to manage these new technologies. The
the size and number of components. A representa- second problem is the initial bottom-up approach used
tive example is the retrieval of photos of all possible by the firm to implement AI projects as it reduces
switch configurations, which requires a huge amount the standardization/replicability of the project on other
of time and internal resources to provide all the sites.
Figure 2. Summary of the main results and discussions.
The main barrier encountered by Company E is the 5.1. Answering RQ1: how do AI applications support
limited availability of economic resources for invest- supply chain and operations management
ment. Another major barrier is the sector’s particularly processes?
advanced technologies, a consequence of the lack of ade-
Using the SCOR classification, most of the projects (13
quate capital to invest in supporting technological inno-
out of 17) involved make processes, with a focus on man-
vation. For example, the company finds it difficult to
ufacturing, quality, and maintenance applications. Two
move into the area of predictive maintenance because
projects involved the plan process, with a particular focus
its suppliers of production equipment have not, to date,
on demand planning and forecasting and inventory man-
offered this type of service included in machinery, which
agement applications, and two involved the deliver pro-
would require sensors and data analytics systems to iden-
cess, with a particular focus on process enquiry and quote
tify deviations in plant behaviour and predict problems
applications. The focus on make processes can be com-
before they occur.
mented on by observing the concentration of expected
Company F has identified three main issues in the
benefits, detailed in the next paragraph, on the dimen-
development of AI projects. The first is the creation of
sions of time and cost reduction, and improvement of
a data-driven company culture, which is important for
production efficiency. In this phase of the life cycle of
the implementation of a suitable IT infrastructure for col-
AI technologies, the need to justify investments made
lecting, storing, preparing, and analysing data. Another
by seeking tangible benefits through their implementa-
critical issue is the ability to decide which AI project to
tion may be predominant. From this perspective, any
develop because the investments are extremely high and
improvement in production processes can generate faster
take a long time to implement, and subsequently gener-
returns on the resources invested.
ate an uncertain value and ROI. The last critical issue is
Almost all the applications developed for quality con-
finding people who have the skills and abilities to imple-
trol are based on image-processing solutions. Despite
ment such technologies. It is often necessary to turn to
the critical issues highlighted by some authors regarding
external consultancies that can support, at least in the
the performance of these systems in particular environ-
initial stages, the development of the project and pro-
mental conditions (Dai et al. 2021), the cases analysed
vide the appropriate skills that are not available within
show how companies have identified in the image anal-
the company.
ysis solutions, particularly effective tools for searching
and classifying defects, with the aim of preventing non-
conformities from ending up in the market, and improv-
5. Discussion ing the production process due to a more accurate analy-
This study empirically investigates 17 AI projects imple- sis of the defects themselves (for example, any recurrence
mented by six companies. For each SCOR process, the AI of specific types of defects, of parts of the product where
projects adopted to support the supply chain and opera- the defects are concentrated, etc.).
tions processes were defined, thus providing an answer to Applications related to manufacturing are more var-
RQ1. The benefits achieved for each AI project were then ied, but one that has come up in more cases is the analysis
reported, providing an answer to RQ2. Finally, the main of variables related to the behaviour of systems to con-
barriers encountered by companies in implementing AI stantly monitor the process and solve production prob-
were classified to provide answers to RQ3. A summary is lems detected in real time. Applications for predictive
depicted in Figure 2. maintenance are all related and concern the estimation of
the health status and lifetime of a machine, prediction of within companies, three distinct sets emerge: barriers
failures, and classification of their causes. All these appli- that concern the more technical aspects of the implemen-
cations are enabled by the context of smart manufactur- tation processes, barriers related to the more strategic-
ing, where the vertical integration of the company allows organisational aspects, and barriers related to financial
heterogeneous data from the shop floor to be collected aspects.
in real time through sensors and merged and handled in Focusing on the more technical dimensions, the sam-
MES and ERP systems. ple of companies investigated highlighted the complexity
inherent in the process of collecting the data necessary
5.2. Answering RQ2: what are the benefits achieved to feed the algorithms underlying the implemented AI
by companies when using AI to support supply chain solutions. The necessary condition for maximising the
and operations management? learning process of ML and DL systems is the availabil-
ity of large quantities of high-quality data. The more the
As already mentioned in the previous paragraph, AI
data available, the more frequently they are updated, and
projects that have been implemented by companies have
the more differentiated forms they come in, the better
had an impact on reducing costs and improving the
the ability of the algorithm to learn and produce accu-
efficiency of production processes, confirming what has
rate outputs. Five companies substantially confirm what
emerged in recently published studies (see Table 1). Con-
emerges from the analysis of the main contributions that
sidering that multiple responses were possible regarding
have appeared in the literature, highlighting the issue of
the performance achieved by the projects, cost reduc-
data quality as the main barrier to implementation due to
tion, lead time reduction, and OEE were the ones mostly
a low ability to govern the data acquisition processes and
covered by the cases analysed, achieving concrete bene-
the subsequent archiving and processing phases. They
fits on these three dimensions. Most of the answers on
underlined the difficulty of building a sufficiently con-
the improvements highlighted a direct impact in terms
sistent and reliable database to feed the learning process
of lower production costs or efficiency gains. Another
of AI algorithms, and the criticality of the data acquisi-
significant set of responses was related to the improve-
tion process correlated with the constraints posed by the
ment of service level and quality, to increase external
machines currently in use.
performance perceived by the customer and safety, to
Concerning the strategic-organisational dimension,
determine a positive impact on the working conditions of
five companies involved in the research reported a lack
the operators, an objective not so obvious, and, above all,
of adequate technical skills to support the development
not always a priority in company policies. The residual
and implementation of AI solutions as another signifi-
part of the answers consists of improvements concerning
cant problem. This aspect is a particularly critical issue
process reliability, environmental sustainability, and the
because it is evident that today, we are facing a struc-
stock rotation index.
tural deficit of figures with a technical background that
It may be interesting to highlight the processes in
respond to the concrete needs of companies that intend
which the benefits achieved are mainly concentrated.
to start AI projects. From the interviews, it emerges that
There is a clear concentration in the make process, espe-
the path of upskilling and reskilling of figures already
cially regarding quality applications, for which improve-
present in the company appears to be impractical today,
ments have been obtained both in terms of product and
as the skills to be developed are very specific, and for
process quality and efficiency. These results are consis-
the management of these projects, it is necessary to
tent with the adoption of image processing solutions, in
have already acquired a certain amount of experience
terms of reducing costs and increasing the efficiency of
because the problems to be faced are often poorly codi-
the quality control processes themselves. Maintenance
fied. To complete these reflections, it should also be noted
and manufacturing applications also have several ben-
that the AI phenomenon is relatively recent regarding
efits. Maintenance benefits are concentrated on process
business applications; therefore, several companies have
efficiency and improvement by reducing lead times and
not yet developed the models and tools required to iden-
costs and improving OEE. Manufacturing applications
tify and select the necessary AI skills. Additional evidence
have various effects, with benefits focused on safety, envi-
emerged from the cases is identified in the absence of
ronment, and process reliability.
an adequate culture of change that can support organisa-
tional structures and individuals in the adoption of tech-
5.3. Answering RQ3: what are the barriers tackled
nological solutions that can also have significant impacts
by companies when implementing AI to support
on processes and areas of comfort that have been created
supply chain and operations management?
over the years. However, there is also an issue of lack of
From the analysis of the barriers that can hinder or readiness from operators, low integration and collabora-
slow down the process of implementing AI solutions tion among organisational departments and supply chain
partners, difficult process changes, and ineffective change processes and specifically the SCOR processes. The
management. results of the literature review show applications of AI in
In terms of financial aspects, four companies high- the five SCOR processes, identify benefits for companies
lighted the financial dimension as a significant barrier in OSCM, and analyse the main barriers to implemen-
to AI adoption. They reported that both the amount tation that companies may face during the implementa-
of investment required to develop and implement the tion of Industry 4.0, which includes AI among its tech-
projects and the difficulty in determining the potential nologies. Very few studies have empirically tackled these
return on investment with a reasonable degree of reli- issues before, although recent studies have emphasised
ability. These results also confirm the indications that the importance of exploring the topic and deepening and
emerged in several studies that reported a lack of financial refining the literature (Nikseresht et al. 2022; Wilson,
resources to support the necessary investments and scep- Paschen, and Pitt 2021). Previous studies have primar-
ticism about achieving a sufficient return on investment ily focused on the development of new AI algorithms,
from AI solutions as obstacles. This last aspect can repre- testing them via simulations or single experimental cases,
sent an important break for those companies that require or proposing literature reviews on the topic without any
very precise analyses of the expected returns to authorise empirical data.
investments, and that embrace a very short-term culture The discussion of the findings of this study shows
for which actions with certain and immediate impacts are how the cases provide new observations, build on
preferred, even if not particularly consistent under the the literature, and introduce interesting new concepts
economic profile. that will advance the theory and practice of AI in
A final consideration relates to the absence of evidence OSCM. The potential applications and benefits identi-
on two factors which often find specific findings in the fied in the literature have been enriched by the inclu-
literature (see Table 2). The companies investigated did sion of real companies’ perspectives. The analysed com-
not highlight any problems on front of the commitment panies concentrated their AI applications to support
and buy-in of the top management. Usually, in such large- the make processes, with a focus on manufacturing,
scale projects, as well as in lean transformation projects or quality, and maintenance applications. These applica-
ERP system implementation projects, these elements are tions made it possible for companies to achieve con-
often cited as the causes of failure. Similarly, no specific crete benefits in three performance dimensions: cost
evidence has emerged on the IT security front associ- reduction, lead time reduction, and OEE. In addition,
ated with the adoption of AI solutions. However, since this study analysed barriers that concern AI implemen-
the analysed companies represent successful cases in the tation, which has resulted in very little coverage by
implementation of AI projects, they have probably over- recent studies, and only some technical barriers related
come these criticalities, thanks to a series of prerequisites to the development of specific ML algorithms have
already satisfied, that is, top management commitment been identified. The results of the case research, start-
and cyber security contracts with suppliers. Rather, they ing from barriers related to Industry 4.0 implementation,
have highlighted only the most relevant barriers still rep- which includes AI among its technologies, show that
resenting real complex obstacles to be overcome, even for the key barriers faced by the companies analysed are
future projects. related to technical, strategic-organisational, and finan-
cial aspects.
6. Conclusion
6.2. Contribution to practice
The aim of this study is to investigate the applications
of AI to OSCM through the experiences and insights From an empirical perspective, this study reports and
of real industrial cases. The study aims to analyse how analyses several projects of AI application to OSCM
AI applications support OSCM in various specific pro- processes within real companies. At present, practition-
cesses and to investigate benefits achieved and barriers ers can find very few models or indications based on
faced by companies in their implementation. To do so, empirical data in the literature. This study fills this gap
this research conducts multiple case studies. It analyses and provides practitioners with practical examples of AI
empirical data concerning 17 applications of AI to the applications to OSCM using a process-based perspective.
SCOR processes within six Italian companies. This perspective can inspire and guide pilot projects on
targeted processes, such as those that create bottlenecks
or those that contribute most to value creation. In addi-
6.1. Contribution to theory
tion, the results of this study can also make managers and
From a theoretical perspective, this study increases the practitioners aware of the possible challenges and benefits
understanding of AI applications to support OSCM that can be encountered during the implementation and
the performance to be assessed to verify the effectiveness Notes on contributors

of the projects. Violetta Giada Cannas, Ph.D. is an Assis-
tant Professor at the School of Industrial
6.3. Limitations and future research Engineering of Carlo Cattaneo Univer-
sity (LIUC). She teaches Operations man-
The main limitation of this study lies in the nature of the agement, Quality management and Smart
methodology adopted. Although multiple case studies Factory at LIUC, and Operations Man-
agement at the School of Management of
are broadly acknowledged as a robust research method Politecnico di Milano. She is also part of
for theory building (Voss, Johnson, and Godsell 2016; the team that work in the i-FAB LIUC, a learning factory for
Robert K Yin 2018), it is also true that it is a qualitative the adoption of Industry 4.0 technologies in a lean-managed
methodology based on a limited set of companies, which assembly system. Her research interests include supply chain
makes it possible to deeply analyse cases but difficult to and production management, lean management, and Indus-
try 4.0, with a particular focus on companies operating in the
generalise the results obtained. Moreover, in this study,
engineer-to-order context.
the cases were all carried out in the same geographical
context, i.e. Italy. The study adopted various strategies to Maria Pia Ciano, Ph.D. is Assistant Pro-
fessor in Operations Management at the
support validity and reliability (see section 3.4), but cau-
Business School of the University of Not-
tion must be exercised in generalising the results of this tingham. She teaches modules on Logis-
study to a larger population. tics and Supply Chain Management, Tech-
The results of this study could be validated by future nology and Organisation, and Plant Loca-
research by replicating this study with a different sam- tion and Design. She completed her PhD
ple or by conducting a quantitative study on a larger at LIUC Carlo Cattaneo University with
a thesis titled ‘Towards an industrial transformation: the link
population. between Lean techniques and Industry 4.0 technologies’. Her
Despite its limitations, the study provides interesting main research topics and interests are: Lean Manufacturing and
and novel results, presenting clear insights into the use Supply Chain, Industry 4.0, Industry 5.0, and the research areas
of AI in various processes of the SCOR model. Future that can derive from or link these topics, such as Data Analytics
studies could further explore the use of AI in ‘return’ and Circular Economy.
processes. They could also take inspiration from the Raffaele Secchi is Full Professor of Oper-
recent developments of the Association for Supply Chain ations and Supply Chain Management at
Management (ASCM), which extends the processes view LIUC – Università Cattaneo, Castellanza.
He is also Dean of LIUC Business School.
given by the SCOR with one based on Digital Capa- He is Visiting Professor at IESEG Business
bilities, summarised in the Digital Capabilities Model School. He has authored or co-authored
(DCM). According to the ASCM, in the DCM ‘each digi- several articles, book chapters, and four
tal capability is mapped to relevant elements in the SCOR books on operations and supply chain
Digital Standard’1 ï£ĳ. This possible future research direc- management. His research appeared in International Journal of
Operations and Production Management, International Jour-
tion would integrate the two views to provide a com-
nal of Production Economics, Industrial Marketing Manage-
prehensive guide to the implementation of AI in OSCM. ment and Production Planning & Control.
Studying the processes to understand the capabilities
Mattia Saltalamacchia is a teaching assis-
required is fundamental and this study provided the basis
tant at the School of Economics of Carlo
for this. Cattaneo University (LIUC). Currently he
is a PhD student under the supervision of
Prof. Raffaele Secchi at the same School.
Note His main topics of interest and research
1. https://www.ascm.org/corporate-transformation/dcm/ are: Lean Management, Operations Man-
agement, Artificial Intelligence, Customer
Journey, Industry 4.0, and performance improvement.
Acknowledgements
We would like to express our gratitude to the Italian Ministry Data availability statement
of Universities and Research (MUR) for their funding support The data that support the findings of this study are avail-
to the author Violetta Giada Cannas through the PON research able from the corresponding author, Maria Pia Ciano (maria.
contract on innovation, enabling technologies, and the broader [email protected]), upon reasonable request.
theme of digitalization.
References
Disclosure statement Abbas Naqvi, Mohammad, and Saman Hassanzadeh Amin.
No potential conflict of interest was reported by the authors. 2021. “Supplier Selection and Order Allocation: A Literature
Review.” Journal of Data, Information and Management, Cantini, Alessandra, Mirco Peron, Filippo De Carlo, and Fabio
125–139. https://doi.org/10.1007/s42488-021-00049-z. Sgarbossa. 2022. “A Decision Support System for Configur-
Agostini, Lara, and Roberto Filippini. 2019. “Organizational ing Spare Parts Supply Chains Considering Different Manu-
and Managerial Challenges in the Path Toward Industry facturing Technologies.” International Journal of Production
4.0.” European Journal of Innovation Management 22 (3): Research, 1–21. https://doi.org/10.1080/00207543.2022.204
406–421. https://doi.org/10.1108/EJIM-02-2018-0030. 1757
Aktepe, Adnan, Emre Yanık, and Süleyman Ersöz. 2021. Carrillo-Hermosilla, Javier, Pablo del Río, and Totti Kön-
“Demand Forecasting Application with Regression and Arti- nölä. 2010. “Diversity of Eco-Innovations: Reflections from
ficial Intelligence Methods in a Construction Machinery Selected Case Studies.” Journal of Cleaner Production 18
Company.” Journal of Intelligent Manufacturing 32 (6): (10-11): 1073–1083. https://doi.org/10.1016/j.jclepro.2010.
1587–1604. https://doi.org/10.1007/s10845-021-01737-8 02.014.
APICS. 2017. Supply Chain Operations Reference Model Version Carvalho, Thyago P., Fabrízzio A.A.M.N. Soares, Roberto
12.0. http://www.emeraldinsight.com/doi/10.110809576059 Vita, Roberto da P. Francisco, João P. Basto, and Symone
710815716. G.S. Alcalá. 2019. “A Systematic Literature Review of
Bahadori, Mohammadkarim, Seyed Morteza Hosseini, Ehsan Machine Learning Methods Applied to Predictive Mainte-
Teymourzadeh, Ramin Ravangard, Mehdi Raadabadi, and nance.” Computers & Industrial Engineering 137 (Novem-
Khalil Alimohammadzadeh. 2020. “A Supplier Selection ber): 106024. https://doi.org/10.1016/j.cie.2019.
Model for Hospitals Using a Combination of Artificial Neu- 106024.
ral Network and Fuzzy VIKOR.” International Journal of Chan, Felix T.S., Anuj Prakash, and Nishikant Mishra. 2013.
Healthcare Management 13 (4): 286–294. https://doi.org/10. “Priority-Based Scheduling in Flexible System Using AIS
1080/20479700.2017.1404730. with FLC Approach.” International Journal of Production
Bai, Chunguang, Patrick Dallasega, Guido Orzes, and Joseph Research 51 (16): 4880–4895. https://doi.org/10.1080/002075
Sarkis. 2020. “Industry 4.0 Technologies Assessment: A Sus- 43.2013.774500.
tainability Perspective.” International Journal of Production Chauhan, Chetna, Amol Singh, and Sunil Luthra. 2021.
Economics 229: 107776. https://doi.org/10.1016/j.ijpe.2020. “Barriers to Industry 4.0 Adoption and Its Performance
107776 Implications: An Empirical Investigation of Emerging
Bala, P. K. 2012. “Improving Inventory Performance with Clus- Economy.” Journal of Cleaner Production 285: 124809.
tering Based Demand Forecasts.” Journal of Modelling in https://doi.org/10.1016/j.jclepro.2020.124809
Management 7 (1): 23–37. Chehbi-Gamoura, Samia, Ridha Derrouiche, David Damand,
Baryannis, George, Sahar Validi, Samir Dani, and Grigoris and Marc Barth. 2020. “Insights from Big Data Analytics
Antoniou. 2019. “Supply Chain Risk Management and Arti- in Supply Chain Management: An All-Inclusive Literature
ficial Intelligence: State of the Art and Future Research Review Using the SCOR Model.” Production Planning and
Directions.” International Journal of Production Research 57 Control 31 (5): 355–382. https://doi.org/10.1080/09537287.
(7): 2179–2202. https://doi.org/10.1080/00207543.2018.15 2019.1639839.
30476 Chen, Ming Kuen, Teng Wang Tai, and Tsu Yi Hung.
Belhadi, Amine, Sachin Kamble, Samuel Fosso Wamba, 2012. “Component Selection System for Green Supply
and Maciel M. Queiroz. 2022. “Building Supply-Chain Chain.” Expert Systems with Applications 39 (5): 5687–5701.
Resilience: An Artificial Intelligence-Based Technique and https://doi.org/10.1016/j.eswa.2011.11.102.
Decision-Making Framework.” International Journal of Pro- Cugno, Monica, Rebecca Castagnoli, and Giacomo Büchi.
duction Research 60 (14): 4487–4507. https://doi.org/10.1080 2021. “Openness to Industry 4.0 and Performance: The
/00207543.2021.1950935. Impact of Barriers and Incentives.” Technological Forecasting
Benitez, Guilherme Brittes, Néstor Fabián Ayala, and Ale- and Social Change 168: 120756. https://doi.org/10.1016/j.tec
jandro G Frank. 2020. “Industry 4.0 Innovation Ecosys- hfore.2021.120756
tems: An Evolutionary Perspective on Value Cocreation.” Dai, W., D. Li, D. Tang, Q. Jiang, D. Wang, H. Wang, and Y.
International Journal of Production Economics 228: 107735. Peng. 2021. “Deep Learning Assisted Vision Inspection of
https://doi.org/10.1016/j.ijpe.2020.107735 Resistance Spot Welds.” Journal of Manufacturing Processes
Birkel, Hendrik, and Julian M Müller. 2021. “Potentials of 62: 262–274. https://doi.org/10.1016/j.jmapro.2020.12.015.
Industry 4.0 for Supply Chain Management Within the Dhamija, Pavitra, and Surajit Bag. 2020. “Role of Artifi-
Triple Bottom Line of Sustainability–A Systematic Litera- cial Intelligence in Operations Environment: A Review
ture Review.” Journal of Cleaner Production 289: 125612. and Bibliometric Analysis.” TQM Journal 32 (4): 869–896.
https://doi.org/10.1016/j.jclepro.2020.125612 https://doi.org/10.1108/TQM-10-2019-0243.
Blome, Constantin, and Tobias Schoenherr. 2011. “Supply Dosdoğru, Ayşe Tuğba, Aslı Boru İpek, and Mustafa Göçken.
Chain Risk Management in Financial Crises—A Multiple 2021. “A Novel Hybrid Artificial Intelligence-Based Deci-
Case-Study Approach.” International Journal of Production sion Support Framework to Predict Lead Time.” Interna-
Economics 134 (1): 43–57. https://doi.org/10.1016/j.ijpe.20 tional Journal of Logistics Research and Applications 24 (3):
11.01.002. 261–279. https://doi.org/10.1080/13675567.2020.1749249.
Cadden, Trevor, Denis Dennehy, Matti Mantymaki, and Ray- Dubey, Rameshwar, Angappa Gunasekaran, Stephen J. Childe,
mond Treacy. 2022. “Understanding the Influential and David J. Bryde, Mihalis Giannakis, Cyril Foropon, David
Mediating Role of Cultural Enablers of AI Integration to Roubaud, and Benjamin T. Hazen. 2020. “Big Data Analyt-
Supply Chain.” International Journal of Production Research ics and Artificial Intelligence Pathway to Operational Per-
60 (14): 4592–4620. https://doi.org/10.1080/00207543.2021. formance Under the Effects of Entrepreneurial Orientation
1946614 and Environmental Dynamism: A Study of Manufacturing
Organisations.” International Journal of Production Eco- Hamdan, Sadeque, and Anwar Jarndal. 2017. A Two Stage
nomics 226 (December 2019): 107599. https://doi.org/10.10 Green Supplier Selection and Order Allocation Using AHP and
16/j.ijpe.2019.107599. Multi-Objective Genetic Algorithm Optimization. https://doi.
Duriau, Vincent J., Rhonda K. Reger, and Michael D. Pfarrer. org/10.1109/ICMSAO.2017.7934843.
2007. “A Content Analysis of the Content Analysis Literature Hays, Patricia A. 2004. “Case Study Research.” In Founda-
in Organization Studies: Research Themes, Data Sources, tions for Research: Methods of Inquiry in Education and the
and Methodological Refinements.” Organizational Research Social Sciences, edited by Kathleen deMarrais, and Stephen
Methods 10 (1): 5–34. https://doi.org/10.1177/10944281062 D. Lapan, 217–234. Mahwah, New Jersey: Lawrence Erl-
89252. baum Associates, Inc.
Dutta, Pankaj, Sahil Talaulikar, Vinay Xavier, and Shubham Helo, Petri, and Yuqiuge Hao. 2021. “Artificial Intelligence in
Kapoor. 2021. “Fostering Reverse Logistics in India by Operations Management and Supply Chain Management:
Prominent Barrier Identification and Strategy Implementa- An Exploratory Case Study.” Production Planning and Con-
tion to Promote Circular Economy.” Journal of Cleaner Pro- trol 0 (0): 1–18. https://doi.org/10.1080/09537287.2021.188
duction 294: 126241. https://doi.org/10.1016/j.jclepro.2021. 2690.
126241 Hill, Alessandro, and Jürgen W. Böse. 2016. “A Decision Sup-
Eisenhardt, Kathleen M. 1989. “Building Theories from Case port System for Improved Resource Planning and Truck
Study Research.” Academy of Management Review 14 (4): Routing at Logistic Nodes.” Information Technology and
532–550. https://doi.org/10.2307/258557. Management 18: 241–251. https://doi.org/10.1007/s10799-
Ellefsen, Anna Paula Tanajura, Joanna Oleśków-Szłapka, Grze- 016-0267-3.
gorz Pawłowski, and Adrianna Toboła. 2019. “No Title.” Log- Islam, Samiul, Saman Hassanzadeh Amin, and Leslie J. Ward-
forum 15 (3): 363–376. https://doi.org/10.17270/J.LOG.20 ley. 2021. “Machine Learning and Optimization Models for
19.354. Supplier Selection and Order Allocation Planning.” Inter-
Ellram, Lisa M. 1996. “The Use of the Case Study Method national Journal of Production Economics 242 (December):
in Logistics Research.” Journal of Business Logistics 17 (3): 108315. https://doi.org/10.1016/j.ijpe.2021.108315.
93–138. Jabbour, Charbel José Chiappetta, Angelo Saturnino Neto,
Encapera, Angelo, Abhijit Gosavi, and Susan L. Murray. José Alcides Gobbo, Maisa de Souza Ribeiro, and Ana
2021. “Total Productive Maintenance of Make-to-Stock Beatriz Lopes de Sousa Jabbour. 2015. “‘Eco-Innovations
Production-Inventory Systems via Artificial-Intelligence- in More Sustainable Supply Chains for a Low-Carbon
Based ISMART.” International Journal of Systems Science: Economy: A Multiple Case Study of Human Critical Suc-
Operations & Logistics 8 (2): 154–166. https://doi.org/10.10 cess Factors in Brazilian Leading Companies.” Interna-
80/23302674.2019.1707906. tional Journal of Production Economics 164 (June): 245–257.
Fallahpour, Alireza, Yew Wong Kuan, Udoncy Olugu Ezu- https://doi.org/10.1016/j.ijpe.2014.11.015.
tah, and Siti Nurmaya Musa. 2017. “A Predictive Inte- Kamble, Sachin S, Angappa Gunasekaran, Abhijeet Ghadge,
grated Genetic-Based Model for Supplier Evaluation and and Rakesh Raut. 2020. “A Performance Measurement Sys-
Selection.” International Journal of Fuzzy Systems 19), tem for Industry 4.0 Enabled Smart Manufacturing Sys-
https://doi.org/10.1007/s40815-017-0324-z. tem in SMMEs-A Review and Empirical Investigation.”
Fashoto, Stephen Gbenga, Akinnuwesi Boluwaji, Owolabi Olu- International Journal of Production Economics 229: 107853.
mide, and Adelekan David. 2016. “Decision Support Model https://doi.org/10.1016/j.ijpe.2020.107853
for Supplier Selection in Healthcare Service Delivery Using Kantasa-ard, Anirut, Maroua Nouiri, Abdelghani Bekrar,
Analytical Hierarchy Process and Artificial Neural Net- Abdessamad Ait el cadi, and Yves Sallez. 2021. “Machine
work.” African Journal of Business Management 10 (9): Learning for Demand Forecasting in the Physical Internet:
209–232. https://doi.org/10.5897/AJBM2016.8030. A Case Study of Agricultural Products in Thailand.” Inter-
Fosso Wamba, Samuel, Maciel M. Queiroz, Cameron Guthrie, national Journal of Production Research 59 (24): 7491–7515.
and Ashley Braganza. 2021. “Industry Experiences of Artifi- https://doi.org/10.1080/00207543.2020.1844332.
cial Intelligence (AI): Benefits and Challenges in Operations Kaparthi, Shashidhar, and Daniel Bumblauskas. 2020. “Design-
and Supply Chain Management.” Production Planning and ing Predictive Maintenance Systems Using Decision Tree-
Control 0 (0): 1–13. https://doi.org/10.1080/09537287.2021. Based Machine Learning Techniques.” International Jour-
1882695. nal of Quality & Reliability Management 37 (4): 659–686.
Frank, Alejandro Germán, Lucas Santos Dalenogare, and https://doi.org/10.1108/IJQRM-04-2019-0131.
Néstor Fabián Ayala. 2019. “Industry 4.0 Technologies: Khashei, Mehdi, and Fatemeh Chahkoutahi. 2021. “Electric-
Implementation Patterns in Manufacturing Companies.” ity Demand Forecasting Using Fuzzy Hybrid Intelligence-
International Journal of Production Economics 210: 15–26. Based Seasonal Models.” Journal of Modelling in Man-
https://doi.org/10.1016/j.ijpe.2019.01.004 agement 17 (1): 154–178. https://doi.org/10.1108/JM2-06-
Gegovska, T., R. Köker, and T. Çakar. 2020. “Green Supplier 2020-0159.
Selection Using Fuzzy Multiple-Criteria Decision-Making Kim, Young-In, and Hyun-Jung Kim. 2021. “Rescheduling
Methods and Artificial Neural Networks.” Computational of Unrelated Parallel Machines with Job-Dependent Setup
Intelligence and Neuroscience. https://doi.org/10.1155/2020/ Times under Forecasted Machine Breakdown.” Interna-
8811834. tional Journal of Production Research 59 (17): 5236–5258.
Guo, Xuesong, Zhiping Yuan, and Bojing Tian. 2009. “Sup- https://doi.org/10.1080/00207543.2020.1775910.
plier Selection Based on Hierarchical Potential Support Kolomvatsos, Kostas, Kyriaki Panagidi, Ioannis Neokosmidis,
Vector Machine.” Expert Systems with Applications 36 (3): Dimitris Varoutas, and Stathes Hadjiefthymiades. 2016.
6978–6985. https://doi.org/10.1016/j.eswa.2008.08.074. “Automated Concurrent Negotiations: An Artificial Bee
Colony Approach.” Electronic Commerce Research and Appli- (February): 294–309. https://doi.org/10.1016/j.matcom.20
cations 19 (September): 56–69. https://doi.org/10.1016/j.eler 18.08.011.
ap.2016.09.002 Luo, Bo, Haoting Wang, Hongqi Liu, Bin Li, and Fangyu Peng.
Kumar, Prashant, and Ananda Shankar Hati. 2021. “Review 2019. “Early Fault Detection of Machine Tools Based on
on Machine Learning Algorithm Based Fault Detection in Deep Learning and Dynamic Identification.” IEEE Transac-
Induction Motors.” Archives of Computational Methods in tions on Industrial Electronics 66 (1): 509–518. https://doi.
Engineering 28 (3): 1929–1940. https://doi.org/10.1007/s11 org/10.1109/TIE.2018.2807414.
831-020-09446-w. Magaldi, Danielle, and Matthew Berler. 2020. “Semi-Structured
Kumar, Veepan, Prem Vrat, and Ravi Shankar. 2021. “Priori- Interviews.” Encyclopedia of Personality and Individual Dif-
tization of Strategies to Overcome the Barriers in Industry ferences, 4825–4830. https://doi.org/10.1007/978-3-319-24
4.0: A Hybrid MCDM Approach.” Opsearch 58 (3): 711–750. 612-3_857
https://doi.org/10.1007/s12597-020-00505-1 Majumdar, Abhijit, Himanshu Garg, and Rohan Jain. 2021.
Kuo, R. J., Y. C. Wang, and F. C. Tien. 2010. “Integration of “Managing the Barriers of Industry 4.0 Adoption and Imple-
Artificial Neural Network and MADA Methods for Green mentation in Textile and Clothing Industry: Interpretive
Supplier Selection.” Journal of Cleaner Production 18 (12): Structural Model and Triple Helix Framework.” Computers
1161–1170. https://doi.org/10.1016/j.jclepro.2010.03.020. in Industry 125: 103372. https://doi.org/10.1016/j.compind.
Lăzăroiu, G., T. Kliestik, and A. Novak. 2021. “Internet of 2020.103372
Things Smart Devices, Industrial Artificial Intelligence, and Mani, Annapoorni, Shahriman Abu Bakar, Pranesh Krish-
Real-Time Sensor Networks in Sustainable Cyber-Physical nan, and Sazali Yaacob. 2021. “Categorization of Mate-
Production Systems.” Journal of Self-Governance and Man- rial Quality Using a Model-Free Reinforcement Learning
agement Economics 6 (1): 20. https://doi.org/10.22381/jsme9 Algorithm.” Journal of Physics: Conference Series 2107 (1):
120212. 012027. https://doi.org/10.1088/1742-6596/2107/1/012027.
Lee, Jay, Hossein Davari, Jaskaran Singh, and Vibhor Pandhare. Manimuthu, Arunmozhi, V. G. Venkatesh, V. Raja Sreedharan,
2018. “Industrial Artificial Intelligence for Industry 4.0- and Venkatesh Mani. 2021. “Modelling and Analysis of Arti-
Based Manufacturing Systems.” Manufacturing Letters 18 ficial Intelligence for Commercial Vehicle Assembly Process
(October): 20–23. https://doi.org/10.1016/J.MFGLET.2018. in VUCA World: A Case Study.” International Journal of Pro-
09.002. duction Research April: 1–19. https://doi.org/10.1080/00207
Lei, Deming. 2010. “A Genetic Algorithm for Flexible Job 543.2021.1910361.
Shop Scheduling with Fuzzy Processing Time.” Interna- Miles, Matthew B., and A. Michael Huberman. 1994. Qual-
tional Journal of Production Research 48 (10): 2995–3013. itative Data Analyses: An Expanded Sourcebook. 2nd ed.
https://doi.org/10.1080/00207540902814348. Thousand Oaks, CA: SAGE Publications.
Leung, K. H., C. C. Luk, K. L. Choy, H. Y. Lam, and Carman K Mohapatra, P., Lyes Benyoucef, and M. K. Tiwari. 2013. “Real-
M Lee. 2019. “International Journal of Production Research ising Process Planning and Scheduling Integration Through
A B2B Flexible Pricing Decision Support System for Manag- Adaptive Setup Planning.” International Journal of Produc-
ing the Request for Quotation Process Under E-Commerce tion Research 51 (8): 2301–2323. https://doi.org/10.1080/002
Business Environment A B2B Flexible Pricing Decision Sup- 07543.2012.715770.
port System for Managing the Request for Qu.” Interna- Naz, Farheen, Rohit Agrawal, Anil Kumar, Angappa
tional Journal of Production Research 57 (20): 6528–6551. Gunasekaran, Abhijit Majumdar, and Sunil Luthra. 2022.
https://doi.org/10.1080/00207543.2019.1566674. “Reviewing the Applications of Artificial Intelligence in Sus-
Li, Dan, Chunhong Liu, and Kang Li. 2019. “A Remanufac- tainable Supply Chains: Exploring Research Propositions for
turing Logistics Network Model Based on Improved Multi- Future directions.” Business Strategy and the Environment 31
Objective Ant Colony Optimization.” Journal Européen Des (5): 2400–2423. https://doi.org/10.1002/bse.3034.
Systèmes Automatisés 52 (4): 391–395. https://doi.org/10.18 Nikseresht, Ali, Bahman Hajipour, Nima Pishva, and Hos-
280/jesa.520409 sein Abbasian Mohammadi. 2022. “Using Artificial Intelli-
Li, Qiubo, and Ru Xiao. 2021. “The Use of Data Min- gence to Make Sustainable Development Decisions Consid-
ing Technology in Agricultural E-Commerce under the ering VUCA: A Systematic Literature Review and Bibliomet-
Background of 6G Internet of Things Communication.” ric Analysis.” Environmental Science and Pollution Research
https://doi.org/10.1007/s13198-021-01108-9. 29 (28): 42509–42538. https://doi.org/10.1007/s11356-022-
Lolli, F., E. Balugani, A. Ishizaka, R. Gamberini, B. Rim- 19863-y.
ini, and A. Regattieri. 2019. “Machine Learning for Multi- Nimawat, Dheeraj, and B. D. Gidwani. 2021. “Identification of
Criteria Inventory Classification Applied to Intermittent Cause and Effect Relationships among Barriers of Industry
Demand.” Production Planning and Control 30 (1): 76–89. 4.0 Using Decision-Making Trial and Evaluation Labora-
https://doi.org/10.1080/09537287.2018.1525506. tory Method.” Benchmarking: An International Journal 28
Loske, Dominic, and Matthias Klumpp. 2021. “Human-AI (8): 2407–2431. https://doi.org/10.1108/BIJ-08-2020-0429.
Collaboration in Route Planning: An Empirical Efficiency- Noman, Abdulla All, Umma Habiba Akter, Tahmid Hasan
Based Analysis in Retail Logistics.” International Jour- Pranto, and A. K. M. Bahalul. 2022. “Machine Learning
nal of Production Economics 241 (November): 108236. and Artificial Intelligence in Circular Economy: A Biblio-
https://doi.org/10.1016/j.ijpe.2021.108236. metric Analysis and Systematic.” Annals of Emerging Tech-
Luan, Jing, Zhong Yao, Futao Zhao, and Xin Song. 2019. “A nologies in Computing (AETiC) 6 (2): 13–40. https://doi.org/
Novel Method to Solve Supplier Selection Problem: Hybrid 10.33166/AETiC.2022.02.002.
Algorithm of Genetic Algorithm and Ant Colony Opti- Nowak, Maciek, Mike Hewitt, and Chelsea C. White. 2012.
mization.” Mathematics and Computers in Simulation 156 “Precedence Constrained Pickup and Delivery with Split
Loads.” International Journal of Logistics Research and Appli- on the Development of a Qualitative Research Instru-
cations 15 (1): 1–14. https://doi.org/10.1080/13675567.2012. ment in Educational Studies.” IOSR Journal of Research
660476. & Method in Education (IOSR-JRME) 12 (1): 22–29.
Olan, Femi, Shaofeng Liu, Jana Suklan, Uchitha Jayawick- https://doi.org/10.9790/7388-1201052229.
rama, and Emmanuel Ogiemwonyi Arakpogun. 2022. “The Sharma, Ajit. 2023. “Making Electric Vehicle Batteries Safer
Role of Artificial Intelligence Networks in Sustainable Sup- Through Better Inspection Using Artificial Intelligence and
ply Chain Finance for Food and Drink Industry.” Interna- Cobots.” International Journal of Production Research, 1–20.
tional Journal of Production Research 60 (14): 4418–4433. https://doi.org/10.1080/00207543.2023.2180308
https://doi.org/10.1080/00207543.2021.1915510. Sharma, Rohit, Anjali Shishodia, Angappa Gunasekaran,
Omwando, Thomas A, Wilkistar A Otieno, Sajjad Farahani, Hokey Min, and Ziaul Haque Munim. 2022. “The Role
and Anthony D Ross. 2018. “A Bi-Level Fuzzy Analytical of Artificial Intelligence in Supply Chain Management:
Decision Support Tool for Assessing Product Remanufac- Mapping the Territory.” International Journal of Production
turability.” Journal of Cleaner Production 174: 1534–1549. Research, https://doi.org/10.1080/00207543.2022.2029611.
https://doi.org/10.1016/j.jclepro.2017.11.050 Sharma, Sugandha, and Gunwant Singh. 2021. “Intelligent
Osterrieder, Philipp, Lukas Budde, and Thomas Friedli. 2020. Warehouse Stocking Using Machine Learning; Intelligent
“The Smart Factory as a Key Construct of Industry 4.0: A Warehouse Stocking Using Machine Learning.” 2021 IEEE
Systematic Literature Review.” International Journal of Pro- International Conference on Mobile Networks and Wireless
duction Economics 221: 107476. https://doi.org/10.1016/j.ij Communications (ICMNWC). https://doi.org/10.1109/ICM
pe.2019.08.011 NWC52512.2021.9688530.
Pasi, Bhaveshkumar N, Subhash K Mahajan, and Santosh Shumanov, Michael, and Lester Johnson. 2021. “Making Con-
B Rane. 2020. “Smart Supply Chain Management: A versations with Chatbots More Personalized.” Computers in
Perspective of Industry 4.0.” Supply Chain Management 29 Human Behavior 117 (April): 106627. https://doi.org/10.10
(5): 3016–3030. 16/j.chb.2020.106627.
Patton, Michael Quinn. 2002. “Two Decades of Develop- Singh, Lakhwinder Pal, and Ravi Teja Challa. 2016. “Integrated
ments in Qualitative Inquiry.” Qualitative Social Work 1 (3): Forecasting Using the Discrete Wavelet Theory and Artifi-
261–283. https://doi.org/10.1177/1473325002001003636. cial Intelligence Techniques to Reduce the Bullwhip Effect in
Pournader, Mehrdokht, Hadi Ghaderi, Amir Hassanzadegan, a Supply Chain.” Global Journal of Flexible Systems Manage-
and Behnam Fahimnia. 2021. “Artificial Intelligence Appli- ment 17 (2): 157–169. https://doi.org/10.1007/s40171-015-
cations in Supply Chain Management.” International Journal 0115-z
of Production Economics 241 (July): 108250. https://doi.org/ Slack, Nigel, Stuart Chambers, and Robert Johnston. 2010.
10.1016/j.ijpe.2021.108250. Operations Management. Harlow, England: Pearson educa-
Pozzi, Rossella, Tommaso Rossi, and Raffaele Secchi. 2021. tion.
“Industry 4.0 Technologies: Critical Success Factors for Stentoft, Jan, Kent Adsbøll Wickstrøm, Kristian Philipsen, and
Implementation and Improvements in Manufacturing Com- Anders Haug. 2021. “Drivers and Barriers for Industry 4.0
panies.” Production Planning & Control 34 (2): 139–158. Readiness and Practice: Empirical Evidence from Small and
https://doi.org/10.1080/09537287.2021.1891481. Medium-Sized Manufacturers.” Production Planning & Con-
Priore, Paolo, Borja Ponte, Rafael Rosillo, and David de la trol 32 (10): 811–828. https://doi.org/10.1080/09537287.20
Fuente. 2019. “Applying Machine Learning to the Dynamic 20.1768318
Selection of Replenishment Policies in Fast-Changing Sup- Strategic Program on Artificial Intelligence. 2022–2024. Jointly
ply Chain Environments.” International Journal of Produc- developed by the The Ministry of Education, University and
tion Research 57 (11): 3663–3677. https://doi.org/10.1080/00 Research, the Ministry of Economic Development, and the
207543.2018.1552369. Minister of Technological Innovation and Digital Transition
Priyadarshi, Rahul, Akash Panigrahi, Srikanta Routroy, and (2021). https://assets.innovazione.gov.it/1637777513-strateg
Girish Kant Garg. 2019. “Demand Forecasting at Retail ic-program-aiweb.pdf.
Stage for Selected Vegetables: A Performance Analysis.” Stuart, I., D. McCutcheon, R. Handfield, R. McLachlin, and D.
Journal of Modelling in Management 14 (4): 1042–1063. Samson. 2002. “Effective Case Research in Operations Man-
https://doi.org/10.1108/JM2-11-2018-0192. agement: A Process Perspective.” Journal of Operations Man-
Raj, Alok, Gourav Dwivedi, Ankit Sharma, Ana Beatriz Lopes agement, https://doi.org/10.1016/S0272-6963(02)00022-0.
de Sousa Jabbour, and Sonu Rajak. 2020. “Barriers to the Svoboda, Josef, and Stefan Minner. 2022. “Tailoring Inven-
Adoption of Industry 4.0 Technologies in the Manufac- tory Classification to Industry Applications: The Benefits of
turing Sector: An Inter-Country Comparative Perspective.” Understandable Machine Learning.” International Journal of
International Journal of Production Economics 224: 107546. Production Research 60 (1): 388–401. https://doi.org/10.10
https://doi.org/10.1016/j.ijpe.2019.107546 80/00207543.2021.1959078.
Riahi, Youssra, Tarik Saikouk, Angappa Gunasekaran, and Tirkolaee, Erfan Babaee, Saeid Sadeghi, Farzaneh Mansoori
Ismail Badraoui. 2021. “Artificial Intelligence Applications Mooseloo, Hadi Rezaei Vandchali, and Samira Aeini.
in Supply Chain: A Descriptive Bibliometric Analysis and 2021. “Application of Machine Learning in Supply Chain
Future Research Directions.” Expert Systems with Appli- Management: A Comprehensive Overview of the Main
cations 173: 114702. https://doi.org/10.1016/j.eswa.2021.11 Areas.” Mathematical Problems in Engineering 2021: 1–14.
4702 https://doi.org/10.1155/2021/1476043.
Ruslin, Ruslin, Saepudin Mashuri, Muhammad Sarib Abdul Toorajipour, Reza, Vahid Sohrabpour, Ali Nazarpour, Pej-
Rasak, Firdiansyah Alhabsyi, and Hijrah Syam. 2022. vak Oghazi, and Maria Fischl. 2021. “Artificial Intelligence
“Semi-Structured Interview: A Methodological Reflection in Supply Chain Management: A Systematic Literature
Review.” Journal of Business Research 122 (January): 502–517. Sustainability 12 (9): 3760. https://doi.org/10.3390/su1209
https://doi.org/10.1016/j.jbusres.2020.09.009. 3760.
Tutun, Salih, Ali Tosyali, Hossein Sangrody, Mohammad Kha- Wu, Cheng-Hung, Fang-Yi Zhou, Chi-Kang Tsai, Cheng-Juei
sawneh, Marina Johnson, Abdullah Albizri, and Antoine Yu, and Stéphane Dauzère-Pérès. 2020. “A Deep Learn-
Harfouche. 2022. “Artificial Intelligence in Energy Indus- ing Approach for the Dynamic Dispatching of Unreli-
try: Forecasting Electricity Consumption Through Cohort able Machines in Re-Entrant Production Systems.” Inter-
Intelligence & Adaptive Neural Fuzzy Inference System.” national Journal of Production Research 58 (9): 2822–2840.
Journal of Business Analytics. Taylor & Francis 6 (1): 59–76. https://doi.org/10.1080/00207543.2020.1727041.
https://doi.org/10.1080/2573234X.2022.2046514. Wuest, Thorsten, Christopher Irgens, and Klaus-Dieter
Unajan, Magdalene C., Bobby D. Gerardo, and Ruji P. Med- Thoben. 2014. “An Approach to Monitoring Quality in Man-
ina. 2019. “A Modified Otsu-Based Image Segmentation ufacturing Using Supervised Machine Learning on Prod-
Algorithm (OBISA).” Lecture Notes in Engineering and Com- uct State Data.” Journal of Intelligent Manufacturing 25 (5):
puter Science 2239 (1): 363–366. 1167–1180. https://doi.org/10.1007/s10845-013-0761-y
Villalba-Diez, Javier, Daniel Schmidt, Roman Gevers, Joaquín Yeh, Wei Chang, and Mei Chi Chuang. 2011. “Using Multi-
Ordieres-Meré, Martin Buchwitz, and Wanja Wellbrock. Objective Genetic Algorithm for Partner Selection in Green
2019. “Deep Learning for Industrial Computer Vision Qual- Supply Chain Problems.” Expert Systems with Applica-
ity Control in the Printing Industry 4.0.” https://doi.org/10. tions 38 (4): 4244–4253. https://doi.org/10.1016/j.eswa.2010.
3390/s19183987. 09.091.
Voss, Chris, Mark Johnson, and Jan Godsell. 2016. “5 Case Yin, R. K. 2009. Case Study Research: Design and Methods. 4th
Research.” Research Methods for Operations Management. ed. Thousand Oaks, CA: SAGE.
Routledge, 165. Yin, Robert K.. 2018. Case Study Research and Applications:
Voss, Chris, Nikos Tsikriktsis, and Mark Frohlich. 2002. “Case Design and Methods. 6th ed. Thousand Oaks, CA: Sage.
Research in Operations Management.” International Journal Zdravković, Milan, Hervé Panetto, and Georg Weichhart. 2022.
of Operations & Production Management 22 (2): 195–219. “AI-Enabled Enterprise Information Systems for Manufac-
https://doi.org/10.1108/01443570210414329. turing.” Enterprise Information Systems 16 (4): 668–720.
Wilson, Matthew, Jeannette Paschen, and Leyland Pitt. 2021. https://doi.org/10.1080/17517575.2021.1941275.
“The Circular Economy Meets Artificial Intelligence (AI): Zeba, Gordana, Marina Dabic, Mirjana Cicak, and Tugrul
Understanding the Opportunities of AI for Reverse Logis- Daim. 2020. “Artificial Intelligence in Manufacturing: Biblio-
tics.” Management of Environmental Quality: An Interna- metric and Content Analysis.” 2020 IEEE / ITU International
tional Journal 33 (1): 9–25. https://doi.org/10.1108/MEQ- Conference on Artificial Intelligence for Good, AI4G 2020,
10-2020-0222. 95–100. https://doi.org/10.1109/AI4G50087.2020.9311087.
Wong, Lai-Wan, Garry Wei-Han Tan, Keng-Boon Ooi, Binshan Zhang, Hao, and Yan Cui. 2019. “A Model Combining a
Lin, and Yogesh K. Dwivedi. 2022. ““Artificial Intelligence- Bayesian Network with a Modified Genetic Algorithm for
Driven Risk Management for Enhancing Supply Chain Green Supplier Selection.” Modeling and Simulation Interna-
Agility: A Deep-Learning-Based Dual-Stage PLS-SEM- tional 95 (12): 1165–1183. https://doi.org/10.1177/0037549
ANN Analysis.” International Journal of Production Rese- 719826306.
arch, 1–21. https://doi.org/10.1080/00207543.2022.2063 Zhao, Xueqing, Min Zhang, and Junjun Zhang. 2021. “Ensem-
089. ble Learning-Based CNN for Textile Fabric Defects Classifi-
Woschank, Manuel, Erwin Rauch, and Helmut Zsifkovits. 2020. cation.” International Journal of Clothing Science and Tech-
“A Review of Further Directions for Artificial Intelligence, nology 33 (4): 664–678. https://doi.org/10.1108/IJCST-12-
Machine Learning, and Deep Learning in Smart Logistics.” 2019-0188.
Appendix 1: case study protocol
Source 1. Face-to-face interview

General questions Company introduction (turnover, employees, sector, product portfolio, business model)
Interviewee/s introduction (role in the company, main interests, experience)
AI applications What are the Artificial Intelligence (AI) applications implemented by your company?
AI and SCOR model • How do AI applications support each of the following supply chain and operations management processes?
• Plan: processes aimed at gathering requirements and information on available resources, balancing requirements
and resources to determine planned capabilities and gaps in demand or resources, and identifying actions to correct
these gaps
• Source: processes aimed at issuing purchase orders, scheduling deliveries, receiving orders, validating orders,
storing goods, and accepting suppliers’ invoices’
• Make: processes aimed at processing, maintenance, repair, overhaul, recycling, refurbishment, manufacturing and
other common types of material-conversion processes
• Deliver: processes aimed at receiving, validating, and creating customer orders, scheduling order deliveries, picking,
packing and shipping, and invoicing customers
• Return: processes aimed at moving material from a customer back through the supply chain to address product
defects, ordering, or manufacturing, or to perform upkeep activities
Benefits and Barriers What benefits have been achieved in implementing and adopting AI applications in the above-mentioned processes?
How many and what barriers have been encountered in implementing and adopting the AI applications in the above-mentioned
processes?
Source 2. Direct observations and internal documents
Company tour Direct observation of the AI applications, with the possibility to ask further questions to the employees and/or managers involved.
Digital or paper materials Internal documents related to the AI projects, with the possibility to ask further questions to the employees and/or managers
involved
Source 3. Official documents
Company’s website General company information; product catalogues and information brochures related to the solutions provided by the companies.
News and press Up-to-date news related to the companies and their AI applications.
National database Economic reports and balance sheets.

AI Report2

Uploaded by

Copyright:

Available Formats

AI Report2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

AI Report2

Uploaded by

Copyright:

Available Formats

International Journal of Production Research

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/tprs20

Artificial intelligence in supply chain and

To link to this article: https://doi.org/10.1080/00207543.2023.2232050

© 2023 The Author(s). Published by Informa

Published online: 12 Jul 2023.

Submit your article to this journal

Article views: 12008

View related articles

View Crossmark data

Citing articles: 4 View citing articles

Full Terms & Conditions of access and use can be found at

Artificial intelligence in supply chain and operations management: a multiple

ABSTRACT ARTICLE HISTORY

1. Introduction (Wong et al. 2022), resilience (Belhadi et al. 2022), and

Table 1. Summary of the literature review on AI applications and beneﬁts.

Table 2. Summary of the literature review on barriers to Industry 4.0 implementation.

Figure 1. Summary of the research steps.

Table 3. Case study overview.

Table 4. Results on AI applications in OSCM.

Table 5. Results on beneﬁts of AI applications in OSCM.

Figure 2. Summary of the main results and discussions.

the performance to be assessed to verify the effectiveness Notes on contributors

Appendix 1: case study protocol

Source 1. Face-to-face interview

You might also like