Received: 31 December 2022 Revised: 19 October 2023 Accepted: 2 January 2024

DOI: 10.1002/bse.3689

RESEARCH ARTICLE

Artificial intelligence and corporate carbon neutrality: A

qualitative exploration

Adeel Luqman 1 | Qingyu Zhang 1 | Shalini Talwar 2 | Meena Bhatia 3 |

4,5,6
Amandeep Dhir

1
Research Institute of Business Analytics and
Supply Chain Management, College of Abstract
Management, Shenzhen University, Shenzhen,
Many firms have established formal carbon neutrality (CN) targets in response to the
China
2
Department of Finance and Economics, S P
increasing climate risk and related regulatory requirements. Subsequently, they have
Jain Institute of Management and Research implemented various measures and adopted multiple approaches to attain these
(SPJIMR), Mumbai, India
3
goals. Academic research has given due attention to firms' efforts in this direction.
Birla Institute of Management Technology,
Greater Noida, India However, past studies have primarily focused on non-digital and process-oriented
4
Department of Management, School of approaches to achieving CN, with the potential of digital technologies such as artifi-
Business and Law, University of Agder,
cial intelligence (AI) remaining less explored. Our study aims to address this gap by
Kristiansand, Norway
5
Jaipuria Institute of Management, Noida, qualitatively examining the use of AI for pursuing CN, drawing insights from firms
India with prior experience in the area. We analyzed the collected qualitative data to iden-
6
Optentia Research Focus Area, North-West
tify four key dimensions that capture different nuances of applying AI for achieving
University, Vanderbijlpark, South Africa
CN: (a) implementing AI for direct and indirect control of emissions, (b) accepting the
Correspondence
strategic trade-offs related to funding, data and systems concerns, and social priori-
Amandeep Dhir, Department of Management,
School of Business and Law, University of ties, (c) overcoming organizational and human-related impediments, and
Agder, Postboks 422 4604 Kristiansand,
(d) acknowledging the significant impact of AI in terms of gains in business model
Norway.
Email: [email protected] efficiency and measurable CN target attainment, which ultimately contribute to
Qingyu Zhang, Research Institute of Business CN. Based on our findings, we propose a convergence–divergence model encom-
Analytics and Supply Chain Management,
College of Management, Shenzhen University,
passing the positive aspects, inhibiting factors, synergies, and offsets necessary for
Shenzhen, China. firms to leverage AI to achieve net-zero emissions effectively. Overall, our study con-
Email: [email protected]
tributes to the discourse on the utilization of AI for CN in a comprehensive manner.
Funding information
This research was supported by Key Project of KEYWORDS
National Social Science Foundation of China artificial intelligence, carbon emission, carbon neutrality, digital technologies, energy
(21AGL014); Natural Science Foundation of management, regulatory mandates
Guangdong—Guangdong Basic and Applied
Basic Research Foundation
(2021A1515011894); Shenzhen Science and
Technology Program
(JCYJ20210324093208022); Shenzhen
University Humanities and Social Sciences
High-level Innovation Team Project for
Leading Scholars (24LJXZ06).

Abbreviations: AI, artificial intelligence; CN, carbon neutrality; CO2, carbon dioxide; EV, electric vehicles; GHG, greenhouse gas; PHEVs, plug-in hybrid electric vehicles; TEMC, transport, energy,
manufacturing, and construction; WEF, World Economic Forum.

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2024 The Authors. Business Strategy and The Environment published by ERP Environment and John Wiley & Sons Ltd.

3986 wileyonlinelibrary.com/journal/bse Bus Strat Env. 2024;33:3986–4003.

 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
LUQMAN ET AL. 3987

[Correction added on 09 February 2024, after

first online publication: Qingyu Zhang was
added as co-corresponding author in this
version.]
[Correction added on 22 March 2024, after
first online publication: Funding Information
has been added in this version.]

1 | I N T RO DU CT I O N emissions of some large corporations (Minevich, 2021). AI is generally

being applied to counter several climate change challenges through
Firms have been actively addressing environmental concerns by actions such as optimizing electricity grids (Chasan, 2023). A deeper
restructuring their manufacturing and distribution practices (Cheng probing of the literature indicates that despite the documented positive
et al., 2021). Concurrently, they have sought to raise awareness by outcomes, the use of AI to achieve CN is not without complexities. It is
launching advertising campaigns promoting the production of eco- generally known that its integration into business activities presents
friendly products such as recycled plastic toothbrushes, toilet paper, various climate-related challenges, including increased energy consump-
and plastic rugs (Zhao et al., 2022), and solar-powered transportation tion, increased greenhouse gas (GHG) emissions through e-waste and
(Okuyama et al., 2022) in addition to bolstering their green reputation recycling processes (Gupta et al., 2021), and issues related to the limited
among consumers (Dolge & Blumberga, 2021). Over time, national lifespan and management of AI waste (M. Fan et al., 2022).
regulators, the United Nations, and scholars have recognized the pro- Broadly speaking, scholars caution that adopting AI comes with lim-
gress made by firms across diverse sectors in reducing carbon emis- itations that should be considered to avoid overestimating its power
sions. However, these endeavors have proven insufficient in enabling (Allal-Chérif et al., 2021). Seeking to examine the overtones and conno-
firms to fulfill their commitments to carbon neutrality (CN) effectively tations in this context, we observe that the debate surrounding the neg-
(Zhang et al., 2022), whereby carbon emissions are balanced with off- ative consequences of AI, such as economic impacts (Fan &
setting approaches (IPCC, 2022). Given that the escalating climate risk Friedmann, 2021), social implications (Algarvio, 2021), and environmen-
has become one of the most significant global challenges, the inability tal performance concerns (Lippiatt et al., 2020), is still evolving, with a
to achieve CN poses an issue that cannot be disregarded. Scholars distinct lack of consensus prevailing in these discussions. This indicates
argue that technological innovations, particularly digital advance- a need for further dissecting the perils of using AI as an apparent pana-
ments, can assist firms grappling with this challenge and facilitate a cea for emission-related challenges. Succinctly, the inability of the exist-
successful transition toward CN. ing literature to provide conclusive evidence on this aspect represents a
Endorsing this view, the World Economic Forum (WEF) acknowl- gap in research. We contend that, given the transformative power of AI
edges that adopting digital technologies can enhance resilience to for businesses and the growing criticality of climate risk, such a gap in
global warming, natural disasters, and emissions control, thereby understanding cannot be left unresolved. In addition to this, a closer
assisting firms in combating anthropogenic emissions and achieving investigation of the literature also indicates that while the past studies
CN (WEF, 2022). According to recent reports, the automation capabil- cover various aspects of the implementation of AI for achieving CN, a
ities of AI can streamline repetitive and time-consuming tasks, significant majority of them have focused on AI-based optimization
enabling businesses to reduce global emissions by up to 4% models. In comparison, the literature is noticeably devoid of empirical
(WEF, 2022). Notably, scholars contend that specific digital technolo- evidence elucidating on-the-ground realities of AI implementation and
gies, such as artificial intelligence (AI), can substantially enhance the other practical issues from a managerial perspective. As a result, the
operational efficiency of firms, make logistic processes resource- accumulated knowledge offers limited insights for firms contemplating
efficient, lower emissions, and support the green economy in general using AI to achieve CN targets. We argue that this is a serious knowl-
(e.g., John et al., 2022; Su & Fan, 2019). edge gap that reduces the practical takeaway of the extant findings and
Motivated by the green potential of AI, existing scholarship has should thus be remedied expeditiously. Existing scholarship has also
expended considerable effort in examining the role of AI in activities taken note of the currently narrow understanding of the long-term
that can inhibit carbon emissions. A comprehensive review of the litera- effectiveness and potential unintended consequences of AI implemen-
ture indicates that past studies have examined the application of AI for tation, calling for further research in this area (Shashi, 2022; Walsh
improving business processes (e.g., Damoah et al., 2021), carbon emis- et al., 2020). Putting the preceding discussion in perspective, we reiter-
sion measurement and forecasting (e.g., Scherz et al., 2022), carbon pric- ate that there is a need to evolve a better understanding of the gains
ing (e.g., Ma et al., 2023), energy management (e.g., Zhao et al., 2021a; that AI has to offer and the trade-offs involved in its adoption for the
Zhou et al., 2022b), energy systems (e.g., Diao et al., 2021; Li & achievement of CN targets.
Maréchal, 2023), renewable energy (e.g., Heo et al., 2022; Motivated by this need, we propose to examine and elucidate the
Natgunanathan et al., 2023), transportation scheduling (e.g., Wang actual experiences of firms that have already implemented AI in their
et al., 2022), and waste management (e.g., Hai et al., 2022; Li businesses to achieve CN targets. Formally, we seek to address the
et al., 2022). Proof of the growing use of AI for reducing emissions is following research question (RQ): How have firms leveraged AI in
Boston Consultancy Group's CO2 AI, which has helped boost the accu- their businesses to pursue and achieve their CN targets, and what
racy of emission measurement and enabled a nearly 40% reduction in have been the offsets its adoption entailed for these firms?
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3988 LUQMAN ET AL.

Consistent with our research question, which aims to explore the the use of AI in these and other sectors, such as health care, can be
less understood aspects of utilizing AI for achieving CN, we have categorized under the following clusters, which are distinct but not all
adopted an inductive grounded theory approach introduced by Glas- voluminous in terms of the number of studies: (a) business processes,
ser and Strauss (1967). To operationalize this approach, we collected (b) carbon emissions and measurement, (c) carbon pricing, (d) energy
qualitative data through open-ended essays, allowing us to gain a management, (e) energy systems, (f) renewable energy,
comprehensive understanding of the experiences and perspectives of (g) transportation scheduling, and (h) waste management. The key
firms that have already implemented AI for CN. Analysis of the col- findings under each of the clusters are summarized below.
lected data helped us identify seven dimensions that offer a fine- In discussing the use of AI for some or other business activities/
grained explanation of the intricacies of using AI for pursuing CN: processes, a set of studies has noted its contribution to improved
direct emission control measures, indirect emission control measures, decision-making and reduced environmental consequences. For
strategic trade-offs, organizational impediments, people impediments, instance, Allal-Chérif et al. (2021) discussed how AI was being used
achieving CN targets, and business model efficiencies. effectively in the purchasing departments of firms by taking procure-
The significant contribution of our study comes from the compre- ment activities beyond just operational decisions to make them more
hensiveness of insight it offers on various aspects of the use of AI for strategic, and Su and Fan (2019) discussed the efficacy of AI in green
achieving CN. To elaborate, our findings provide a 360-degree under- logistics. Providing evidence to further support the positive impact of
standing of the topic at hand by explaining the purpose of implemen- AI as a key enabling technology for business processes, John et al.
tation, the unavoidable trade-offs, the impediments that act as (2022) noted its capability to reduce barriers impeding sustainability-
barriers to the effective leveraging of AI for achieving CN, and a wide- related innovations, and Damoah et al. (2021) showcased its effective-
ranging explanation of the positive and negative impact on people and ness in healthcare supply chains.
processes. By providing such a multi-faceted understanding, our study A notable number of recent studies on using AI to achieve some
offers a useful point of reference for managers, clarifying actionable or the other CN objectives have discussed three aspects: carbon emis-
pathways to support their strategic decisions. We further enhance the sions, emission measurement, and carbon pricing. Most studies con-
industry relevance of our study by offering a precise set of factors forming to the two clusters—carbon emissions and measurement—
that make the implementation of AI appealing, and an extensive set of have either emphasized the need for better forecasting of emissions
factors that compel managers to take a more cautious view while con- to support policymaking or suggested ways to measure emissions in
templating the use of AI for achieving CN targets. The comprehensive different sectors. Studies aimed at addressing the need for improved
framework developed inductively also offers a bird's eye view to forecasting have suggested models such as new information-based
researchers aiming to advance the research in the area. gray model (Ding & Zhang, 2023), ensemble prediction system with
In sum, our study contributes to both research and practice by sys- both point and interval estimates of future emissions (Liu et al., 2022),
tematically revealing the convergences and divergences inherent in the and predictive model for estimating global warming (Babatunde
application of AI for achieving CN. At the same time, by analyzing the et al., 2020). In addition, some studies also proposed models for deter-
qualitative responses collected from 22 firms that have already imple- mining fuel combinations and building designs to lower emissions. For
mented AI-based solutions for pursuing their CN targets, we present a example, Bhowmik et al. (2018) developed a model to determine the
detailed and nuanced view of the synergies and offsets that need to be optimal operating parameters for varying fuel combinations and
fully appreciated for the successful implementation of AI for CN. Scherz et al. (2022) presented decision support systems that could be
useful in designing carbon-neutral buildings.
Coming to research on carbon pricing, studies have made specific
2 | A I A N D CN contributions, such as assessment of carbon markets as a solution to
offset the costs of emissions efficiently (Cao et al., 2022), the use of
Climate change and risk are the most debated topics in scholarly litera- optimization techniques for purchasing carbon emission rights in car-
ture, popular press, industry forums, and policy circles. A large part of the bon spot and future markets (Ma et al., 2023), and the development
conversation has now swung from identifying the climate risks to dis- of a multi-factor decomposition and integration carbon price forecast-
cussing how to best combat the issues. CN has taken a central role in this ing model (Zhao et al., 2021b).
discourse, and the approaches that can help to achieve it have become As expected, the largest number of studies on the application of
the focal point of concern. The resultant literature has been focused AI for CN focused on one or more aspects of energy. We divided
largely on several process-based approaches, such as carbon capture and them into three clusters: energy management, energy systems, and
storage, with a comparatively narrow focus on digital technologies to renewable energy. The studies categorized under energy management
achieve CN. Nevertheless, the limited literature that has evolved, particu- discussed aspects such as consumption, efficiency, and optimization.
larly around the use of AI for CN, is quite informative and promising. Of these, the emphasis on the development of predictive and other
A review of the literature on the role of AI in climate change models was quite apparent, with studies proposing models such as a
reveals that existing scholarship has acknowledged the effectiveness fuzzy logic-based model for predicting the concentration of harmful
of AI in achieving CN goals in sectors such as energy, transportation, emissions in flue gases arising from combustion of coal and biomass
construction-building, and manufacturing (e.g., Shi et al., 2021; Xi (Krzywanski et al., 2022), a theoretical intelligent strategic bidding
et al., 2021; Zhou et al., 2022b) (see Table 1). In sum, the research on simulation model for the electricity market (Wu et al., 2022), and a
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
LUQMAN ET AL. 3989

TABLE 1 Literature review.

Study Data and Method Country Main variables

Ding and Zhang (2023) New-information-based gray model/Monte-Carlo China Forecasting provincial carbon emissions
simulation/probability density analysis/
experiments
Ma et al. (2023) Lyapunov optimization technique/theoretical China Carbon-aware ML task offloading/green edge
analysis and simulation/modeling AI/purchasing carbon emission rights in carbon
spot and future markets
Li and Maréchal (2023) Optimization/Monte Carlo simulation/AI/case Switzerland Energy autonomy/carbon neutrality/negative
study emissions/uncertainty impacts on the energy
system
Natgunanathan et al. ML-based prediction models/computational Australia Prediction of generated power from the Deakin
(2023) experiments/real-world data generated from solar farm
Deakin Microgrid
John et al. (2022) Systems thinking/multi-level perspective/review Netherlands Socio-technical impact of AI as a key enabling
of technical scientific articles/semi-structured technology
interviews
Scherz et al. (2022) Hierarchical decision modeling (HDM)/systems UK Integration of sustainability requirements inthe
thinking methodology/life-cycle perspective/ early design phase of buildings/carbon-neutral
modeling buildings
Cao et al. (2022) Metrics for evaluating the carbon efficiency of a China Carbon neutral data centers/carbon market
data center
Krzywanski et al. (2022) Fuzzy logic/simulation/experimental studies Poland Prediction of polluting gaseous emissions from
advanced combustion
Wu et al. (2022) Multi-agent transfer learning (MATL)/multi-agent China Electricity market/MATL algorithm comparison
reinforcement learning (MARL)/simulation
experiment/open source data
Zhou et al. (2022b) Modeling/adaptive neural fuzzy inference UK and Reduce the development workload for the energy
system/Gaussian process regression/ China management controller
experimental evaluations
Heo et al. (2022) Data processing techniques/variational South Korea Forecasting model for offshore wind farms
autoencoder (VAE)/explainable artificial (OWFs)/techno-economic and environmental
intelligence (XAI)/deep learning-based assessments
generative modeling/stochastic scenarios/
offshore wind speed datasets and energy data
obtained from petrochemical industrial parks
Wang et al. (2022) Optimization/modeling/hybrid algorithm based China EV charging stations
on PSO and gray wolf/sensitivity and
robustness analysis
Hai et al. (2022) Genetic algorithm/simulation/sensitivity analysis China Waste heat recovery/GHG emission reduction/
integrated energy system
Huseien et al. (2022) Metaheuristic shuffled frog leaping algorithm/ Singapore Construction material/waste recycling/reduction
experiment in energy consumption
Li et al. (2022) ML algorithms China Predicting CO2 production from green waste
composting
Liu et al. (2022) Time-varying filter-based empirical mode China Ensemble prediction system for forecasting
decomposition (TVF-EMD)/statistics/ANNs/ emissions
deep learning predictors
Zhou et al. (2022a) Review of learning mechanisms of AI-based China AI in buildings and renewable energy systems/
applications/review of AI applications for carbon-neutrality transition in building sector
renewable energy utilization and intelligent
buildings
Zhou (2022b) Review of advanced ocean energy converters/ China Ocean energy/reducing carbon emissions/carbon
power supply characteristics of ocean energy neutrality/application of AI
resources/hybrid ocean energy storages with
synergies/application of diversified ocean
energy systems/application of AI

(Continues)
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3990 LUQMAN ET AL.

TABLE 1 (Continued)

Study Data and Method Country Main variables

Zhao et al. (2021b) Hodrick–Prescott filter/meta-analysis/partial China Carbon price forecasting model
correlation function (PACF)/least absolute
shrinkage and selection operator (LASSO)/
modeling
Zhao et al. (2021a) A variety of battery sizes and innovations for EV China Environmental carbon pollution and
and PHEVs/modeling management/AI assisted V2G for a plug-in
hybrid vehicle
Allal-Chérif et al., 2021) Exploratory, inductive, and qualitative approach France Purchasing and the performance of purchasing
based on a multiple case studies departments
Damoah et al. (2021) Documentary and in-depth semi-structured Ghana AI-enhanced medical drone application in health-
interviews care supply chain
Rocha et al. (2021) Elitist non-dominated sorting genetic algorithm Brazil Energy demand planning in smart homes/forecast
II/support vector regression technique/ of a distributed generation
numerical simulations/real data from a smart
home/K-means clustering technique
Xi et al. (2021) Gradient boosted regression trees with Bayesian China Efficient utilization of gases in steel mills for
optimization/particle swarm optimization (PSO) power supply/deployment of carbon capture
algorithm/case studies/sensitivity analyses/ technologies, use of renewable power/carbon
real-world weather data neutrality
Diao et al. (2021) Convolutional neural networks (DL-CNN) China Performance prediction of power cycle
Liu et al. (2021) Machine learning algorithms China New energy materials/data-driven materials/
global carbon neutrality/application of AI & ML
Shukhobodskiy et al. RED WoLF hybrid storage system/simulation/ UK Energy consumption in residential dwellings/
(2021). progressive threshold approach reducing load from the electrical grid/savings in
emissions
Sadoudi et al. (2021) Elephant herding optimization/scenario analysis Algeria Interconnected microgrid/renewable energy
sources/power management
Babatunde et al. (2020) Case study/predictive modeling/fuel flowrates Nigeria Global warming potential/carbon tax
and air–fuel mass ratios (AFRs)/Aspen HYSYS
8.8/GaBi 8.0/GMDH Shell DS 3.8.9/Microsoft
Excel software
Tien et al. (2020) Convolutional neural network/modeling/building UK Building energy management systems/vision-
energy simulation (BES) tool based deep learning approach
Kempitiya et al. (2020) AI-based bidding optimization/case study Finland Frequency reserves of renewable energy/
frequency reserves market
Kumar et al. (2020) Simulation-optimization approach India Energy consumption of electric-powered bus
system
Zhang et al. (2020) AI and metaheuristic algorithms Australia Optimizing mixtures of recycled aggregate
concrete (RAC)
Su and Fan (2019) Optimization/modeling/computational China Intelligent logistics/green logistics/green
experiments with simulation data vehiclerouting problem
Pulselli et al. (2019) Carbon accounting methodology Italy Assess GHG emissions in urban environments/
carbon footprint of urban neighborhoods/
spatial visualization of forestland
Huang et al. (2019) Optimization/modeling/case study South Korea Industrial energy management
Kumar et al. (2019) Optimization/modeling/economic analysis India Multi-energy off-grid microgrid
Bhowmik et al. (2018) Multi-objective response surface methodology India Performance and exhaust emissions of diesel
(MORSM)/modeling/artificial neural network/ engines fueled by Diesosenol blends
AI

transfer learning routine to support real-time energy management for et al., 2021a), residential buildings (Tien et al., 2020), and off-grid
plug-in hybrid electric vehicles (PHEVs; Zhou et al., 2022b). Several microgrid (Kumar et al., 2019).
other studies developed AI-based models for energy management in Continuing the discussion on energy, but shifting focus to energy
different contexts, such as smart homes (Rocha et al., 2021), systems, we find that the congruent studies in this cluster also
manufacturing (Huang et al., 2019; Xi et al., 2021), PHEVs (Zhao focused on developing AI-based models. For example, X. Li and
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
LUQMAN ET AL. 3991

Maréchal (2023) developed a model for quantifying the impact of At the same time, scholars are still debating whether AI brings with it
uncertainty on national-level energy systems, and Diao et al. (2021) certain economic, social, and other issues (Algarvio, 2021; Fan &
proposed a deep learning architecture for energy system design and Friedmann, 2021). The debate is far from settled and remains too
predicting power cycle performance. coarse-grained at the moment. Overall, the understanding of adverse
Another prominent cluster of studies found in the congruent liter- outcomes of using AI for pursuing CN, various barriers encountered
ature focused on the role of AI in the management of renewable by firms trying to implement AI for this purpose, and the tangible out-
energy. This is as expected since one of the prominent ways of lower- comes of such an implementation are limited.
ing emissions that is considered feasible the world over is the substi-
tution of fossil fuels with renewable energy sources. In some of the
recent studies in this regard, Natgunanathan et al. (2023) discussed 3 | METHODOLOGY
the digital twin of the Deakin Microgrid, a facility for researching solar
energy data, and Zhou (2022b) reviewed the use of AI to understand 3.1 | Research context
its role in promoting sustainable ocean energy systems.
Notably, several studies grouped under this cluster also The research context of our study is situated in the broad socio-
developed and discussed AI-based models such as an explainable AI- economic landscape characterized by escalating climate risk, rising
based generative model producing stochastic scenarios for renewable debate on the responsibility of firms to control polluting emissions,
energy (Heo et al., 2022); an AI-based optimization and control strat- global calls for pursuing CN targets, and strict regulatory mandates
egy for an interconnected microgrid of wind turbines, solar power, imposed by governments. Within this context, our study aims to
and photovoltaic generators (Sadoudi et al., 2021); a generalized investigate the on-the-ground realities and real-life experiences of
model for the operation of frequency reserve markets for renewable firms that have taken the initiative to implement AI in pursuit of their
energy (Kempitiya et al., 2020); and AI applications for CN targets. Furthermore, our study focuses on firms operating in the
renewable energy utilization (Y. Zhou, 2022a). transport, energy, manufacturing, and construction (TEMC) sectors,
The remaining relevant studies either discussed transportation since these are known to be among the highest emissions
scheduling or waste management since these two activities can sub- contributors.
stantially impact emissions and their reduction. In these two clusters
too, the previously observed emphasis on AI-based model development
was noted. The studies related to transportation proposed models such 3.2 | Research design
as optimal and economic EV charging stations (Wang et al., 2022) and a
simulation-optimization approach to manage energy consumption in Although climate change has been a widely discussed topic, the
electric-powered bus systems (Kumar et al., 2020). debate around corporate CN and the use of AI for achieving it is still
In the case of the final cluster, waste management, the models in an evolutionary state. As discussed in the preceding sections, our
proposed include an AI-based genetic algorithm to reduce emissions understanding of various aspects of leveraging AI for CN is not as
and environmental costs of a hybrid energy system based on fuel from refined as it could be. As a result, to better elucidate the intricacies
a wastewater treatment plant (Hai et al., 2022), an informational involved, we opted to undertake an exploratory qualitative examina-
algorithm-based model to assess high-strength alkali-activated mor- tion of the topic. Our decision to employ a qualitative research design
tars made from waste bottle glass nanoparticles with ground blast fur- is motivated by scholarly recommendations based on its inherent
nace slag and fly ash (Huseien et al., 2022), a hybrid intelligent system emphasis on exploring intricate phenomena, capturing rich descrip-
to determine optimal mixtures of recycled coarse aggregate (Zhang tions, and generating comprehensive insights (Maxwell, 2000). Using a
et al., 2020), and an ML-based model to predict CO2 release during qualitative approach facilitated an in-depth exploration of the intrica-
green waste composting (Li et al., 2022). cies and varied dimensions of implementing AI for achieving CN.
The preceding discussion underscores the notion that past stud- Specifically, we used open-ended essays to collect written
ies are skewed towards optimization models and design issues. As a responses from participants through a set of questions. This method
result, empirical insights on the implementational aspects and the has been gaining popularity in the academic literature (e.g., Dhir
firm-level experiences related to the use of AI for pursuing CN targets et al., 2017; Talwar et al., 2021a). The essay-type survey method
are practically non-existent. This gap in understanding severely limits enabled us to seek detailed textual responses from the study respon-
the usefulness of the accumulated findings, making it imperative for dents, providing extensive input for a comprehensive exploration of
researchers to suitably address the deficit. firms' experiences. We derived the set of open-ended questions for
Furthermore, although the literature around the use of AI is the study by comprehensively reviewing the available literature and
heavily skewed towards acknowledging its multiple benefits, some seeking expert input to refine them. Through these questions, we
studies in different settings have raised a word of caution that the tried to understand how firms currently use AI to achieve CN, how
integration of AI in businesses might itself pose climate-related chal- effective those initiatives are, what the different implementation-
lenges arising from energy consumption, e-waste management, and related challenges and problems are, and what major gains, if any,
related environmental issues (M. Fan et al., 2022; Gupta et al., 2021). have emerged.
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3992 LUQMAN ET AL.

3.3 | Data collection and sampling aggregate dimensions. Our choice of the Gioia et al. (2013) approach
was guided by the knowledge that using an established coding method
We solicited participation from employees of firms that had already enhances the credibility and rigor of the data analysis process. We were
implemented AI to achieve CN. The participants were recruited through also motivated by the knowledge that qualitative researchers widely
an online platform, and participation was kept voluntary. Since we adopt the coding method proposed by Gioia et al. (2013) due to its
wanted participation among employees who could be considered key effectiveness in analyzing textual data and generating meaningful
informants based on their knowledge and experience using of AI for CN, insights (Hannigan et al., 2019; Petrescu et al., 2023).
we executed the data collection process in multiple phases or waves, fol- Following this approach to analyzing the data purely inductively,
lowed by data analysis. The entire process is illustrated in Figure 1. we used respondents' words to populate first-order concepts. There-
In the first and the second phases, we conducted two screening after, we systematically developed the second-order themes, consoli-
surveys, inviting participation from full-time employees of firms oper- dating them into aggregate dimensions. We kept the first-order
ating in the TEMC sectors. The purpose of screening surveys was to concepts as close to informants' voices as possible, with the second-
identify key informants who work at managerial levels and are closely order theme being researcher-based.
associated with the AI-CN project in their firms. In the third phase, we Adequate precautions were taken to ensure objectivity and mini-
collected qualitative responses to our main study's open-ended ques- mize confirmation bias in our coding process. We followed a multi-step
tions. We kept the study open, checking the data continuously, and process, wherein, first, two authors independently coded the data to
closed it when theoretical saturation was reached. As a result, we generate zero-order codes, which were then consolidated into first-
ended up with 22 complete responses from respondents working in order concepts. Discrepancies in coding were discussed to achieve
full-time managerial roles in firms in the TEMC sectors. inter-coder reliability. All coding was done manually. In the second
Of the 22 respondents, the majority were male (86%). The age step, the remaining members of the author team, who were not a part
distribution of respondents fell into the following ranges: 26–30 years of the initial coding, checked the codes generated by the coders. They
(18%), 31–35 years (36%), 36–40 years (9%), 41–45 years (23%), and added their comments and observations. In the third step, we com-
51 years or above (23%). In terms of employment, manufacturing had pared the results of both groups to determine the second-order
the highest representation (55%), followed by construction (27%). The themes and aggregate dimensions. Since the responses were solicited
transportation/logistics and energy sectors accounted for approxi- from experienced professionals working in firms with relevant, hands-
mately 14% and 5% of the respondents, respectively. on experience in using AI to achieve CN targets, there was quite a lot
In the fourth phase, we analyzed all 22 responses inductively to of clarity and preciseness in the collected data. As a result, there were
generate first-order concepts, second-order themes, and aggregate very few discrepancies in the coding executed independently by the
dimensions, as described below. author team. While we found minor discrepancies in the naming of
aggregate dimensions, the consensus was easy to achieve.

3.4 | Data analysis

3.5 | Validity and reliability
We followed the iterative coding process suggested by Gioia et al.
(2013) to analyze the data, presenting the findings through a data struc- In qualitative research, ensuring validity and reliability is crucial for
ture diagram comprising first-order concepts, second-order themes, and maintaining the credibility and trustworthiness of the findings.

FIGURE 1 Phases of data collection and analysis.

 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
LUQMAN ET AL. 3993

Following the seminal recommendations of constructive and evalua- The third category, impediments, captures the resistance faced by
tive procedures (e.g., Guba & Lincoln, 1981; Whittemore et al., 2001), firms in their endeavor to use AI for CN. Under this category, we have
as used varyingly in recent studies (e.g., Iyanna et al., 2022), we grouped two aggregate dimensions—organizational impediments and
employed several strategies to enhance the validity and reliability of people impediments. Of these, organizational impediments comprise
our study, as tabulated in Table 2. two second-order themes: financial concerns and operational con-
cerns, and people impediments comprise two second-order themes:
steep learning curve and stakeholder resistance.
4 | RESULTS Finally, the fourth and last category, impact, represents the out-
comes and consequences of implementing AI for CN. We grouped
Through the three-step coding process described above, we devel- two aggregate dimensions—business model efficiencies and achiev-
oped 18 second-order constructs and consolidated them to obtain ing CN targets—under impact, with the former comprising three
seven aggregate dimensions. The data structure diagram is presented second-order themes: stakeholder impact, process impact, and
in Figures 2(a–c). bottom-line impact, and the latter comprising two second-order
For ease of understanding and clarity of explanation, we have fur- themes: measurable emission outcomes and better alignment with
ther categorized the aggregate dimensions under four illustrative CN goals.
heads: implementation, trade-offs, impediments, and impact. Imple-
mentation captures how the firms are using AI for CN. We have
grouped direct emission control measures and indirect emission con- 4.1 | Implementation of AI
trol measures under this head. Herein, direct emission control mea-
sures comprise three second-order themes: emission management 4.1.1 | Direct emission control measures
(measurement and reduction), energy consumption, and sustainability-
oriented practices, and indirect emission control measures comprise The direct impact on emissions dimension captures the AI-based ini-
three second-order themes: leanness, operational efficiency, and func- tiatives of firms introduced with the express purpose of controlling
tional effectiveness. emissions that increase the carbon and other harmful content in the
Next, trade-offs represent the give and take the firms had to con- atmosphere. These activities broadly include actions such as emission
sider in executing their plans to use AI for CN. In our data structure, management through measurement and reduction, tracking and opti-
trade-offs comprise only one aggregate dimension—strategic trade- mizing energy consumption, and supporting sustainability-oriented
offs—with three second-order themes: seeking funding support, addres- practices. Analysis of textual responses indicated that about 70% of
sing data and systems concerns, and acknowledging social priorities. firms were deploying AI for activities that can be considered direct
emission control measures.
TABLE 2 Validity and reliability. One of the control activities is the measurement of emissions
being released. Responses confirmed that firms were explicitly using
Strategy Operationalization Process
AI for measuring GHG emissions arising from their business activity.
Sample Identification of key Two-step
appropriateness informants for screening Aligned with their commitment to act as per their environmental
participating in the survey accountability, these firms were using AI for the daily measurement of
study GHG levels. As a respondent stated:
Prolonged Dedicating sufficient Comprehensive
engagement time to understanding literature “To help measure the amounts of greenhouse gases
the research context, review
released into the atmosphere by our primary oil and natu-
screening the A two-step
participants, screening ral gas drilling, extraction, and refining activities”
collecting the data, survey followed [P3, M, 51 or above, Energy]
and analyzing it by the main
study
In addition to measuring emissions, AI is also being used to reduce
Data analysis
through a series them. In this regard, key approaches used by the firms were tracking
of iterative energy consumption, analyzing where the consumption was not as
steps justified, and optimizing machine settings and production operations
Member checking Designating two Coding of data to avoid wasteful consumption and exhaust. As highlighted by one
authors to respondent,
independently code
the data, followed by
checking the same by “Artificial intelligence has helped us in optimizing the con-
two other authors sumption of electricity at all our facilities. This has helped
who were not part of us prevent and eliminate any leakages”
the initial analysis
[P14, M, 41–45, Construction]
10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
LUQMAN ET AL.

Data structure of strategic trade-offs, organizational impediments, and people impediments.

Data structure of direct and indirect emission control measures.

FIGURE 2b
FIGURE 2a
3994
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
LUQMAN ET AL. 3995

FIGURE 2c Data structure of business model efficiencies and carbon neutrality targets.

Another respondent shared: 4.1.2 | Indirect emission control measures

“We have bespoke, in-house developed AI software which The indirect impact on emissions dimension captures the AI-based ini-
monitors and, in some cases, commences maintenance of tiatives of firms introduced to support activities that would ultimately
machinery. This is a proactive software program and it contribute to controlling emissions, even though the link might not be
ensures machinery is running at optimum performance, apparent. Such initiatives are broad-based, covering a plethora of
thus reducing exhaust to the atmosphere” activities that make the business efficient, less wasteful, and profit-
[P5, F, 51 or above, Manufacturing] able. Analysis of responses helped us identify three distinct activities
undertaken by the surveyed firms that reduced carbon emissions and
To elaborate, firms use AI to control machine operations, turning other harmful content in the atmosphere. These activities broadly
them on only when needed. AI is also used to monitor the amount of include actions taken to enhance leanness, operational efficiency, and
energy being used so that it can prevent energy use when it is not the functional effectiveness of the business as a whole.
needed. The objective is to save as much power and fuel as possible. Analysis of responses confirmed that all surveyed firms had inte-
Furthermore, AI is being used in these firms to support grated AI into their businesses to make them less wasteful, efficient,
sustainability-oriented practices such as reduced reliance on manual and effective. The responses provide insights into how organizations
reporting methods, resulting in decreased paper usage and electronic are utilizing AI to reduce manual monitoring, reporting, and recycling
device consumption, thereby conserving valuable resources. At the methods by empowering machines to operate themselves while learn-
same time, AI is being used to support material waste reduction in ing to do so most efficiently. In addition, the analysis reveals that firms
manufacturing processes and encourage more sustainable practices. are using AI to optimize resource allocation, enhance productivity,
To explain further, AI has been implemented to analyze production minimize errors, and reduce project durations, resulting in reduced
processes, identify opportunities for minimizing material waste, and environmental impact associated with their projects and cost savings.
optimize resource usage. By implementing AI algorithms, firms are try- As a respondent highlighted,
ing to find ways to use fewer materials in their day-to-day operations,
contributing to a decreased environmental footprint and promoting “We have human-assisted machines that learn to sepa-
sustainable practices. The following statement exemplifies this com- rate recyclable materials from non-recyclable ones. These
mitment to resource efficiency and sustainability: machines learn from the instructions of humans and their
mistakes to work autonomously in the future”
“My organization is making use of artificial intelligence [P6, M, 41–45, Construction]
and machine learning to achieve carbon neutrality mostly
to reduce material consumption in manufacturing and to Firms are also using AI algorithms for upgrading production oper-
build different things and also to come up with greener, ations, improving the efficiency of plant balance components, and
more sustainable logistic networks” optimizing logistics and warehousing operations to improve opera-
[P15, F, 30–35, Manufacturing] tional and functional effectiveness. To elaborate, AI helps firms
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3996 LUQMAN ET AL.

optimize the movement of goods, streamline supply chains, and mini- respondent quoted below expressed the entire quandary rather
mize unnecessary material handling and transportation. These succinctly:
advancements indirectly contribute to reducing carbon emissions
associated with logistics operations. As one respondent stated: “I think that if we but aside money, than the biggest
problems and challenges are people who are a bit older
“My organization uses AI to improve and optimize logis- and/or are not so ongoing with all the new technologies
tics and warehousing efficiency, to find a way to use as and changes and are trying to keep things the same way
little materials as possible in day-to-day operations, and as they have always been, because it works, and can't see
all around find ways to reduce carbon emissions as a the bigger picture. Some of them are also afraid that arti-
whole” ficial intelligence will take away everyones jobs”
[P7, M, 21–25, Manufacturing] [P15, F, 31- 35, Manufacturing]

4.2 | Trade-offs 4.3 | Impediments

4.2.1 | Strategic trade-offs 4.3.1 | Organizational impediments

The strategic trade-offs dimension encapsulates the conscious balan- The organizational impediments dimension comprises the barriers or
cing of factors that may accompany the roll-out of their AI implemen- inhibiting factors that temper the enthusiasm and efforts of firms to
tation. These trade-offs encompass considerations related to integrate AI for achieving CN targets. Analysis of responses reveals
governmental interference, data safety issues, and social concerns. two key drivers of organizational impediments: financial concerns and
Analysis of the collected responses helped us identify three compo- operational concerns. The main source of financial concern is that not
nents of such trade-offs: seeking funding support, addressing data only is the initial investment high, but there is also a long gestational
and systems concerns, and acknowledging social priorities. period before the AI project starts delivering on profitability promises.
To elaborate, a significant trade-off identified is the need to This slow crawl to break even can become a grave challenge for firms
accept that AI implementation projects require extensive investment, that are constantly facing topline–bottom-line pressures. Our findings
making it almost unavoidable for firms to seek funding support from indicate that there is a consensus that, despite the substantial upfront
the government. Seeking government funding is more than likely to investment, firms can realize significant long-term financial benefits
bring with it increased scrutiny and various regulatory checks and bal- from using AI for CN. The following quote from a respondent high-
ances. However, given the long recovery period of such investments, lights this view:
firms need to accept government funding as more of a facilitating fac-
tor. Analysis of responses confirmed that this is how firms perceived “The initial expense in developing the in-house software
seeking government funding for implementing AI for CN. In the words was large; however, in the five years since it has been
of a respondent: used, it has certainly paid for itself several times over, not
only in helping the company achieve carbon neutrality
“The main problem is to make it profitable to achieve an but also in maintenance costs”
adequate performance of artificial intelligence in this field [P5, F, 51 or above, Construction]
in a reasonable time, and to obtain subsidies and eco-
nomic aid if necessary from the government to be suc- Another impediment arises from managing a smooth transition to
cessful in the shortest possible time” a new operating system and handling data infrastructure effectively.
[P6, M, 41–45, Construction] Making room for AI entails several adjustments, which are somewhat
challenging to manage. As one respondent admitted,
Another trade-off is the common threat associated with digital
technologies, data security, and the general perception that AI sys- “Biggest issue was integration into current operating sys-
tems' decisions are biased based on the data used to train them. tems as these are shared with our local authority as we
Accordingly, firms using AI for CN have to accept the calculated risk are a sub-contractor organisation”
of potentially biased decisions and data leaks. The final set of strategic [P9, M, 46–50, Transportation]
trade-offs consists of proceeding despite knowing that using AI for
CN might conflict with certain social priorities. The conflict may arise
due to personal privacy risks, which any data-driven setup can be 4.3.2 | People impediments
exposed to. In addition, the respondents acknowledged that there
was a fear of AI replacing humans completely, leading to job loss. Fur- People impediments encompass the man-made challenges and bar-
thermore, a trade-off is required in accepting the impersonal nature of riers associated with using AI for CN. The analysis of qualitative
AI changing social interactions and reducing human interfacing. A responses helped us identify two key manifestations of people's
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
LUQMAN ET AL. 3997

impediment and their causes. One set of such impediments arises To sum up, we see AI, ML as the potential to revelutionise
from the steep learning curve accompanying the new-fangled AI sys- cardon reduction but we are not there quite yet”
tems in terms of acquiring new knowledge to use them initially and [P2, M, 51 or above, Manufacturing]
maintain their consistent performance thereafter. Learning to use the
new systems requires several adjustments and the rather uphill task At the same time, the responses indicate how the use of AI has
of acquiring the relevant skills. One respondent described this impedi- led to a positive impact in terms of streamlining processes, aiding
ment using the following words: informed decision-making, supporting the accurate input of materials
into the process cycle, reducing the time taken to onboard new
“Well, even if we don't have yet this methods implemen- resources, enabling faster project delivery, and generating higher
ted on our company, we have already studied them. We profits. Pinpointing the impact of AI-enabled decision-making in firms,
will certainly face some issues in the begining because AI a respondent opined:
must learn and must be working very well to do a good
job. Maybe we'll need a full time team working exclusively “Modernization technologies are solutions that deal with
on AI, to support it” the sources of emissions from major industrial units that
[P20, M, 31–35, Transportation] are the heart of the industrial process - for example,
replacing an ethylene steam cracker powered by fossil
Another set of people impediments arises from stakeholder resis- fuels with an electric steam cracker”
tance. The staff, as well as top management, may resist the use of AI [P18, M, 41–45, Manufacturing]
for CN. Most of such resistance is due to strong resistance to change
along with a generational divide, with older employees showing more Analysis of responses showed that the use of AI for achieving CN
status quo bias. In the words of a respondent: targets positively impacted the bottom line by offering several cost-
effective benefits and a high return on investment (ROI), albeit after
“Our biggest challenge was to get the buy-in from staff as some time lag. Respondents also noted how AI was effective in help-
it was challenging to change their minds and to get staff ing their firms survive in a highly competitive market. In the words of
and even clients to think differently towards CO2 emis- a respondent:
sions as many felt it's not their responsibility”
[P14, M, 41–45, Construction] “Artificial intelligence helps us so that in the future
machines alone can make decisions and separate mate-
rials without the collaboration of a human, thus saving
4.4 | Impact time and energy. All this will take us a few years to fully
implement and allow intelligence to work autonomously”
4.4.1 | Business model efficiencies [P6, M, 41–45, Construction]

The business model efficiencies dimension embodies the positive

impact of using AI to achieve CN targets on people, processes, and 4.4.2 | Achieving carbon neutrality targets
profits. It emphasizes the potential of AI to enable machines to make
decisions and perform tasks without human collaboration. Analysis of The achieving CN targets dimension captures how effective the
the collected qualitative data helped us capture these efficiencies in a implementation of AI technologies has been in achieving firms'
fine-grained manner through three themes: stakeholder impact, pro- emissions-related goals. We identified two themes within this dimen-
cess impact, and bottom-line impact. These responses provide insights sion: measurable emission outcomes and better alignment with CN
into how AI can enhance business models and drive efficiencies in the goals to lucidly illustrate how using AI for CN delivers the anticipated
TEMC sectors. The theme of stakeholder impact reflects how optimal positive outcomes. About 65% of the respondents indicated that the
outcomes accrue for firms when AI is leveraged to engage different implementation of AI in their firms' day-to-day operations had gar-
stakeholder groups, both internal (e.g., employees) and external nered significant reductions in their carbon footprint and a substantial
(e.g., customers). In the words of a respondent: decrease in emissions so far. These respondents ventured to quote
specific percentages of reduction, as is evident from the response
“In our experience AI technologies and machine learning quoted verbatim:
play a critical role in helping to achive carbon neutrality.
Examples would be tracking assets cars lorrys, electric “They have been quite effective. Numbers have been
devices and wareables via telematics, driving efficencies shown by our statistics team that since the start of AI
of energy, cost, time and money. Other examples include implementation in the organization's day-to-day opera-
workflow management helping employees, customers, tions, we have reduced our carbon footprint by around
and vendors to be targeted when where ans as needed. 60%. The most impactful circumstance is the reduction in
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3998 LUQMAN ET AL.

trips taken by our transportation vehicles. Routes have strengthening its cross-functional and interactive role, and Su and Fan
been immensely optimized so we take as few trips as pos- (2019) showed that major obstacles, such as end distribution in logis-
sible” tics operations, can be overcome, and the entire logistics system can
[P7, M, 21–25, Manufacturing] be made more resource-efficient, cost-effective, and low-carbon
intensive by rationally deploying AI in tandem with big data.
Our findings indicate that the gains in emission reductions are Next, our findings show that firms adopting AI for CN need to
likely to continue since the adoption of AI has made firms better settle for many strategic trade-offs. The essence of trade-offs here is
aligned with CN objectives. This ensures that all the activities of firms that the firms have to proceed with the implementation and contin-
are geared up to reduce emissions or make business efficient in opti- ued usage of AI, knowing there may be some undesirable conse-
mizing decision-making and resource use. The response reproduced quences or at least some serious concerns. Such trade-offs include
below communicates this idea quite unequivocally: dependence on government support, balancing performance and
financial viability, addressing social and data-related concerns,
“By using automation, not only does it reduce human and ensuring long-term sustainability. These trade-offs arise since
error and in some cases human involvement altogether, it upfront costs need to be weighed against long-term benefits such as
ensures accurate input of raw and intermediate materials efficiency, emissions reductions, and safety improvements. This find-
into the process cycle and minimises waste at the end ing is consistent with past studies that have underscored the need to
process too. Monitoring of the systems allows less energy understand the trade-offs, such as loss of accuracy, unknown future
to be used at the same time which also contributes to our costs, and higher emissions associated with AI applications (e.g., Ma
carbon neutrality goal” et al., 2023). On the whole, the past literature has remained bereft of
[P4, M, 51 or above, Manufacturing] a wide-ranging discussion on the complex trade-offs that need to be
accepted by firms desirous of rolling out AI implementation for pursu-
ing CN. Our study addresses this narrowness by presenting an elabo-
5 | DISCUSSION rate set of trade-offs related to governmental interference on account
of providing funding support, data safety issues inherent in such tech-
We sought to answer how firms were leveraging AI in their businesses nologies, and social concerns due to unknown elements that AI
to pursue and achieve their CN targets, and the offsets its adoption encapsulates.
has entailed for them. The findings of our study indicate that the use Coming to impediments, our study brought forth several organi-
of AI for achieving CN by firms can be understood clearly through zational and people impediments related to organizational culture,
four aspects: implementation (i.e., what it is being used for), trade-offs generational divides, resistance to change, top management buy-in,
(i.e., the balancing offsets firms have to accept and deal with), impedi- the need for extensive workforce training, integrating AI systems with
ments (i.e., the barriers that inhibit its effective adoption or use), and existing systems, etc. Although past studies have not delved as deeply
impact (i.e., the outcomes and consequences of implementing AI). into barriers and impediments toward the use of AI for CN, our find-
With regard to implementation, our coding of the qualitative ings are in concordance with the available evidence in the extended
responses revealed that AI was being employed by firms for initiating literature (e.g., Ahmed et al., 2022; Cichosz et al., 2020; Garzoni
and supporting several direct and indirect emission control measures et al., 2020). Our study contributes substantially in this regard
covering activities such as reducing the release of GHGs, measuring because not only are the available insights limited but they also lack
those released, and making efforts to increase the efficiency and the depth and consensus required to guide future research and prac-
effectiveness of businesses. Our findings resonate with past studies tice. To elaborate, while past studies have focused more on climate-
that have emphasized the use of AI for emission management. How- related challenges of using AI (Fan et al., 2022; Gupta et al., 2021), we
ever, our study offers a comprehensive discussion on direct and indi- bring forth a broad gamut of impediments, not only economic but also
rect emission control measures, in comparison to past literature, those related to culture, behavior, and psychology.
which has focused primarily on emissions forecasting models such as Finally, with regard to impact, the coding of textual responses
the new information-based gray model (Ding & Zhang, 2023), predic- helped us confirm that firms were indeed garnering several antici-
tive models capable of estimating global warming (Babatunde pated positive outcomes associated with business model efficiencies
et al., 2020), and ensemble prediction system offering both point and and the achievement of CN targets. These findings are consistent
interval estimates of future emissions (Z. Liu et al., 2022). with and build upon the available insights in the area
Our findings further indicate that organizations actively utilize AI (e.g., Allal-Chérif et al., 2021; Hai et al., 2022; Su & Fan, 2019).
to streamline operations and promote resource conservation, consis- However, where past studies have noticeably discussed the impact
tent with existing literature (e.g., Allal-Chérif et al., 2021; Damoah of AI on CN through the lens of energy systems and management
et al., 2021; Su & Fan, 2019). However, our study covers a broader of renewable energy (e.g., Diao et al., 2021; Li & Maréchal, 2023;
set of avenues compared to the more specific focus of past studies on Natgunanathan et al., 2023), our study presents a more comprehen-
a singular aspect of their choice. For instance, Allal-Chérif et al. (2021) sive perspective spanning business models, in addition to direct dis-
noted AI's contribution to enhancing firms' purchasing function by cussion on CN.
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
LUQMAN ET AL. 3999

5.1 | Conceptual framework results in topline and bottom-line gains, unlocking value by mitigating
climate risks and strengthening the firm's resilience to climate change
Climate risk has become an abiding discussion during the past decade shocks.
(e.g., Kumar et al., 2021), mandating strong action on emissions. Process efficiencies arising from the implementation of AI for CN
Highlighting technology as a potential means of effective action on become apparent through gains in the maintenance schedule, where
emissions, our study confirms that leveraging AI for achieving CN tar- preventive monitoring and predictive maintenance ensure fewer
gets requires a keen understanding and balancing a host of business instances of breakdowns. Similarly, AI aids in optimizing machine set-
and climate-related challenges and outcomes. Our findings underscore tings to automate many processes and material selection to reduce
the idea that bringing together AI, which is a complex technology, and wastage.
CN, which has multiple drivers, is not a simple undertaking. It entails a Climate action, representing the third converging factor in our
deep understanding of various aspects. The existing scholarship has model, captures the apparent and unavoidable outcomes that need to
also acknowledged this, with researchers arguing that it is difficult to be derived from the application of AI for CN. These pertain to emis-
apply any available adoption model to conceptualize and examine the sions reductions, prioritizing the use of renewable energy, monitoring
use of AI for CN (Allal-Chérif et al., 2021). As a result, we bring vehicle movement and logistics to make it more environmentally effi-
together our findings and consolidate them systematically to develop cient, and managing energy consumption in different activities.
a conceptual framework, which we call the convergence–divergence In our model, the diverging factors are also divided into three
model (Figure 3). parts. We conceptualize these challenges as resource constraints,
The intuition behind this model is that the landscape of using AI managerial constraints, and resistance. Resource constraints capture
for CN is a dynamic mix of factors, some of which act to propel the the uncertainties in value generation that arise from the large amount
firms on the path of neutrality and others that detract from the antici- of investment required for AI projects and the long and slow process
pated gains. We conceptualize the first set of factors as converging to break even that becomes even more excruciating due to the
factors. These factors serve as incentives for firms in terms of profit- requirement of specific technical expertise.
ability, environmental performance, and productivity, motivating them The next diverging factor in our model, managerial constraints,
to commit to using AI for achieving CN. embodies the counterbalancing issues arising due to top management
The second set of factors represents the diverging factors, which skepticism, which has a dampening effect on the enthusiasm of the
we conceptualize as the counterbalancing part of our model since entire organization. Managerial issues can also arise due to regulatory
they take away from or temper the positives arising from implement- pressures from the government to implement the AI project in a cer-
ing AI for achieving CN. These factors unleash several organizational tain way, normative pressures from the community to advance social
and people-related challenges that impede the smooth integration of interests, and mimetic pressures driven by a need to ape competitors'
AI in existing systems and potentially delay the gains arising from the actions. Operationally, managerial constraints arise from the extensive
adoption of AI. change management required for transitioning from the existing sys-
We break down the converging factors into three distinct parts: tems to the AI systems and ensuring seamless integration, which is
the use of AI leading to decision optimization, process efficiencies, fraught with challenges.
and commensurate climate action. We contend that decision optimiza- The final diverging factor in our model comes in the form of resis-
tion by implementing AI for the express purpose of achieving CN tance of employees and customers to the adopted AI system, which

FIGURE 3 Convergence–divergence model.

 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4000 LUQMAN ET AL.

might impede the expected efficacious outcomes. Resistance also cap- studies can work on empirically testing the entire framework or some
tures the inhibiting factors that arise from social concerns about AI in parts. For instance, only convergences can be subjected to a more in-
terms of biases, human redundancy in certain roles, and the decline depth examination.
in human interaction. We further include in our model data safety Finally, we provide a concrete direction for future research by
issues and the large amount of data associated with AI systems that offering an efficacious methodological choice in the form of open-
create barriers arising from information paranoia and technostress. ended essays, which are becoming an increasingly popular mode of
The third part of our model captures the path to net zero, which data collection, preferred by scholars working on a wide variety
entails reducing GHG emissions drastically to near zero and absorbing of research topics (e.g., Dhir et al., 2017; Talwar et al., 2021b). Given
emissions, if any remain, from the atmosphere. Through this part, we the complex nature of the area and the evolving state of research,
consolidate our findings to contend that the synergies unlocked by clarity on the research method suitable for the context can add the
converging forces can lead firms toward the state of net zero emis- desired momentum to the research, yielding robust results. At the
sions when they consciously and continuously work towards offset- same time, the open-ended essay protocol of our study provides use-
ting the detracting effect of diverging factors. ful inputs for future empirical studies in the area.

6 | C O N CL U S I O N 6.2 | Practical implications

6.1 | Theoretical implications Our study offers three key practical implications for different stake-
holders: First, our findings reveal that although firms can benefit
Our study offers three valuable theoretical implications: First, we con- immensely from the adoption of AI for achieving their CN targets, the
ducted an exploratory qualitative study through a novel yet provenly high initial investment and the long gestational period of these pro-
efficacious method of open-ended essays to understand and map the jects act as dissuading factors preventing firms from taking any initia-
on-the-ground experiences of firms that have implemented AI with tive. In addition, some unknown costs can arise in the future while
the express purpose of achieving CN. Our analysis uncovered four implementing AI (Ma et al., 2023), which can add to the uncertainty.
dimensions of the AI–CN debate—implementation, strategic trade- Against this backdrop, our findings highlight the criticality of govern-
offs, impediments, and impact. By bringing these four dimensions into ment funding for promoting the use of AI for CN. Given the severity
the same space, we consolidate and systematize the conversations of climate change and the impact of climate risk at a much broader
that have remained anecdotal, dispersed, and fragmented so far. Most level, we recommend that governments institute formal funding pack-
of the existing investigations of AI adoption and its impact on CN ages at reasonable terms, with no restrictive covenants to motivate
have used a narrow lens, focusing either on benefits or on potential firms to adopt AI.
negative outcomes (e.g., Agrawal et al., 2022; Ahmed et al., 2022), Second, our findings show that the resistance of internal and
which has left many aspects less understood, or even unacknow- external stakeholders is one of the major impediments that might pre-
ledged. Scholars have noted this, raising a call for further research in vent firms from adopting AI for CN in the future if they have not done
the area (e.g., Shashi, 2022; Walsh et al., 2020). Our study answers it so far. Even for firms that have adopted AI, the resistive attitude of
this call by providing a multi-dimensional perspective on using AI for different stakeholders can inhibit the successful integration of the
CN. As a result, we have created a suitable knowledge base for future new systems with the old ones. The source of resistance is the new-
research to explore more granular aspects of the implementation of AI ness of the idea of using AI for CN since, currently, in many sectors,
for pursuing and achieving CN targets. AI tools are largely being used for incremental improvements in tech-
Second, we contribute to advancing research in the area by for- nologies rather than nurturing new low-carbon technologies (John
mulating a comprehensive conceptual framework that brings together et al., 2022). Based on this evidence, we recommend that any AI
various convergent and divergent factors impacting AI-related out- implementation project for CN should not only be approached by
comes for firms. Conceptual models presenting a comprehensive firms as a long-term investment decision but also as an organizational
overview of the area under focus are acknowledged for their agenda- process mandating due change management. Furthermore, we sug-
setting contribution to theoretical advancement (e.g., Dhir gest that the resistance of staff can be overcome by altering their
et al., 2020). In addition, we add a strategic perspective to the debate mindset and perception that emissions are not their responsibility.
on the application of AI for CN by revealing that despite several posi- This would require addressing gaps in awareness and cultivating a
tive, well-documented outcomes and synergies of AI, the decision to sense of shared responsibility toward sustainable practices. By foster-
implement AI for achieving CN entails several trade-offs, which need ing such collective understanding and commitment, firms can over-
to be weighed by firms and elucidated by researchers. Furthermore, come people's impediments and promote a culture of sustainability.
not only does our framework advance the theoretical understanding Finally, the majority of respondents indicated that the most chal-
of the area, but it also contributes to promoting industry-oriented lenging issue their firms faced while implementing AI for achieving CN
research by motivating academic researchers to examine critical vari- targets was finding experts to implement this technology and make it
ables related to value, risk, and the top and bottom lines. Future profitable adequately. It is apparent from our findings that AI
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
LUQMAN ET AL. 4001

implementation has been constrained by the limited availability of original draft preparation, and writing—review and editing. Meena
skilled personnel capable of understanding and handling AI from the Bhatia participated in validation and review.
CN perspective. Thus, specialized training is required to develop cli-
mate technology know-how. Offering a potential solution, we recom- ACKNOWLEDG MENTS
mend a multi-pronged approach, spearheaded by deep university– First and second authors are supported by Key Project of National
industry collaboration for training personnel equipped with the Social Science Foundation of China (21AGL014); Natural Science
required skills, competence, and ability to understand how best AI can Foundation of Guangdong—Guangdong Basic and Applied Basic
be leveraged to lead firms toward the state of net zero emissions. Research Foundation (2021A1515011894); Shenzhen Science and
Since most firms lack the readiness and relevant experience to adopt Technology Program (JCYJ20210324093208022); Shenzhen
AI in their business processes (Allal-Chérif et al., 2021), competence- University Humanities and Social Sciences High-level Innovation
building can go a long way toward making firms more receptive to Team Project for Leading Scholars (24LJXZ06).
adopting and effectively leveraging AI technologies. [Correction added on 09 February 2024, after first online publica-
tion: Acknowledgment section was added in this version.]

6.3 | Limitations and future research insights CONFLIC T OF INTER E ST STATEMENT

The authors do not have any competing interests to declare.
Our study offers several tangible contributions, but as with any
research undertaking, it is also not devoid of shortcomings. Since not- OR CID
ing and articulating key limitations of any study also paves the way for Shalini Talwar https://orcid.org/0000-0002-2819-5685
future research, we have given them careful attention. The first limita- Meena Bhatia https://orcid.org/0000-0003-4579-0109
tion is that while at the time of data collection, we invited participa- Amandeep Dhir https://orcid.org/0000-0002-6006-6058
tion from individuals employed in the TEMC sectors only, and we
have not coded the data or reported the result to present sector-wise RE FE RE NCE S
findings. Reporting the results sector-wise could have made the Agrawal, R., Wankhede, V. A., Kumar, A., Luthra, S., & Huisingh, D. (2022).
Progress and trends in integrating Industry 4.0 within circular econ-
insights more granular. Future studies can remedy our limitation by
omy: A comprehensive literature review and future research proposi-
investigating each sector individually to test our findings and highlight tions. Business Strategy and the Environment, 31(1), 559–579. https://
the deviations. In addition, specific sub-sectors that are known to be doi.org/10.1002/bse.2910
highly polluting can be examined, such as the leather industry Ahmed, I., Zhang, Y., Jeon, G., Lin, W., Khosravi, M. R., & Qi, L. (2022). A
blockchain-and artificial intelligence-enabled smart IoT framework for
(Moktadir et al., 2020). Second, we have remained confined to an
sustainable city. International Journal of Intelligent Systems, 37(9),
inductive approach, offering only key themes and a theoretical frame- 6493–6507. https://doi.org/10.1002/int.22852
work. Going beyond this to put forth testable hypotheses, and even Algarvio, H. (2021). The role of local citizen energy communities in the
testing them, could have enhanced the contribution of our study mul- road to carbon-neutral power systems: Outcomes from a case study in
tifold. However, this would have increased the scope of our study. Portugal. Smart Cities, 4(2), 840–863. https://doi.org/10.3390/
smartcities4020043
This limitation opens an opportunity for future researchers to extend
n-Moya, V., & Ballester, A. C. C. (2021). Intelligent
Allal-Chérif, O., Simo
our study by incorporating propositions and testing them with cross- purchasing: How artificial intelligence can redefine the purchasing
sectional data. Finally, we consider a holistic convergence–divergence function. Journal of Business Research, 124, 69–76. https://doi.org/10.
model for mapping all potential factors associated with AI implemen- 1016/j.jbusres.2020.11.050
Babatunde, D. E., Anozie, A. N., Omoleye, J. A., Oyebode, O.,
tation in business operations. However, our conceptualization is not
Babatunde, O. M., & Agboola, O. (2020). Prediction of global warming
grounded in any existing theory, being more closely grounded in the potential and carbon tax of a natural gas-fired plant. Energy Reports, 6,
findings of the qualitative study. This creates scope for researchers to 1061–1070. https://doi.org/10.1016/j.egyr.2020.11.076
identify theories likely to fit our conceptualization. Accordingly, we Bhowmik, S., Paul, A., Panua, R., Ghosh, S. K., & Debroy, D. (2018). Perfor-
mance-exhaust emission prediction of diesosenol fueled diesel engine:
recommend that future research consider landmark theories that have
An ANN coupled MORSM based optimization. Energy, 153, 212–222.
been used effectively in pro-environmental settings. https://doi.org/10.1016/j.energy.2018.04.053
In addition, we suggest that future researchers can build up on Cao, Z., Zhou, X., Hu, H., Wang, Z., & Wen, Y. (2022). Towards a system-
our exploratory study to offer more granular insights on using AI for atic survey for carbon neutral data centers. IEEE Communication Sur-
veys and Tutorials, 24, 895–936. https://doi.org/10.1109/COMST.
achieving CN by considering firm-level differences such as small ver-
2022.3161275
sus large, location in developed versus developing countries, listed Chasan, A. (2023). Some experts see AI as a tool against climate change.
versus privately held firms, and ownership status. Others say its own carbon footprint could be a problem. Available at:
https://www.cbsnews.com/news/artificial-intelligence-carbon-
footprint-climate-change/#::text=But%20experts%20largely%20see
AUTHOR CONTRIBUTION
%20AI,the%20effects%20of%20climate%20change
Adeel Luqman and Qingyu Zhang participated in the conceptualization
Cheng, Y., Sinha, A., Ghosh, V., Sengupta, T., & Luo, H. (2021). Carbon
and writing—original draft preparation. Shalini Talwar and Amandeep tax and energy innovation at crossroads of carbon neutrality:
Dhir participated in methodology, formal analysis, validation, writing— Designing a sustainable decarbonization policy. Journal of
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4002 LUQMAN ET AL.

Environmental Management, 294, 112957. https://doi.org/10.1016/j. considering stochastic scenarios of remotely sensed offshore wind
jenvman.2021.112957 energy. Journal of Cleaner Production, 379, 134793. https://doi.org/10.
Cichosz, M., Wallenburg, C. M., & Knemeyer, A. M. (2020). Digital transfor- 1016/j.jclepro.2022.134793
mation at logistics service providers: Barriers, success factors and lead- Huang, X., Hong, S. H., Yu, M., Ding, Y., & Jiang, J. (2019). Demand
ing practices. The International Journal of Logistics Management, 31(2), response management for industrial facilities: A deep reinforcement
209–238. https://doi.org/10.1108/IJLM-08-2019-0229 learning approach. IEEE Access, 7, 82194–82205. https://doi.org/10.
Damoah, I. S., Ayakwah, A., & Tingbani, I. (2021). Artificial intelligence (AI)- 1109/ACCESS.2019.2924030
enhanced medical drones in the healthcare supply chain (HSC) for sus- Huseien, G. F., Faridmehr, I., Nehdi, M. L., Abadel, A. A., Aiken, T. A., &
tainability development: A case study. Journal of Cleaner Production, Ghoshal, S. (2022). Structure, morphology and compressive strength
328, 129598. https://doi.org/10.1016/j.jclepro.2021.129598 of alkali-activated mortars containing waste bottle glass nanoparticles.
Dhir, A., Chen, G. M., & Chen, S. (2017). Why do we tag photographs on Construction and Building Materials, 342, 128005. https://doi.org/10.
Facebook? Proposing a new gratifications scale. New Media & Society, 1016/j.conbuildmat.2022.128005
19(4), 502–521. https://doi.org/10.1177/1461444815611062 IPCC. (2022). Intergovernmental panel on climate change. https://www.
Dhir, A., Talwar, S., Kaur, P., & Malibari, A. (2020). Food waste in hospital- ipcc.ch/
ity and food services: A systematic literature review and framework Iyanna, S., Kaur, P., Ractham, P., Talwar, S., & Islam, A. N. (2022). Digital
development approach. Journal of Cleaner Production, 270, 122861. transformation of healthcare sector. What is impeding adoption and
https://doi.org/10.1016/j.jclepro.2020.122861 continued usage of technology-driven innovations by end-users. Jour-
Diao, C., Liu, T., Yang, Z., & Duan, Y. (2021). Comparison between deep nal of Business Research, 153, 150–161. https://doi.org/10.1016/j.
learning and fully connected neural network in performance prediction jbusres.2022.08.007
of power cycles: Taking supercritical CO2 Brayton cycle as an example. John, N., Wesseling, J. H., Worrell, E., & Hekkert, M. (2022). How key-
International Journal of Intelligent Systems, 36(12), 7682–7708. https:// enabling technologies' regimes influence sociotechnical transitions:
doi.org/10.1002/int.22603 The impact of artificial intelligence on decarbonization in the steel
Ding, S., & Zhang, H. (2023). Forecasting Chinese provincial CO2 emis- industry. Journal of Cleaner Production, 370, 133624. https://doi.org/
sions: A universal and robust new-information-based grey model. 10.1016/j.jclepro.2022.133624
Energy Economics, 121, 106685. https://doi.org/10.1016/j.eneco. Kempitiya, T., Sierla, S., De Silva, D., Yli-Ojanperä, M., Alahakoon, D., &
2023.106685 Vyatkin, V. (2020). An artificial intelligence framework for bidding opti-
Dolge, K., & Blumberga, D. (2021). Key factors influencing the achieve- mization with uncertainty in multiple frequency reserve markets.
ment of climate neutrality targets in the manufacturing industry: LMDI Applied Energy, 280, 115918. https://doi.org/10.1016/j.apenergy.
decomposition analysis. Energies, 14(23), 8006. https://doi.org/10. 2020.115918
3390/en14238006 Krzywanski, J., Czakiert, T., Nowak, W., Shimizu, T., Zylka, A., Idziak, K.,
Fan, M., Khalique, A., Qalati, S. A., Gillal, F. G., & Gillal, R. G. (2022). Ante- Sosnowski, M., & Grabowska, K. (2022). Gaseous emissions from
cedents of sustainable e-waste disposal behavior: The moderating role advanced CLC and oxyfuel fluidized bed combustion of coal and bio-
of gender. Environmental Science and Pollution Research, 29(14), mass in a complex geometry facility: A comprehensive model. Energy,
20878–20891. https://doi.org/10.1007/s11356-021-17275-y 251, 123896. https://doi.org/10.1016/j.energy.2022.123896
Fan, Z., & Friedmann, S. J. (2021). Low-carbon production of iron and steel: Kumar, A., Sharma, S., & Verma, A. (2019). Optimal sizing and multi-energy
Technology options, economic assessment, and policy. Joule, 5(4), management strategy for PV-biofuel-based off-grid systems. IET Smart
829–862. https://doi.org/10.1016/j.joule.2021.02.018 Grid, 3(1), 83–97.
Garzoni, A., De Turi, I., Secundo, G., & Del Vecchio, P. (2020). Fostering dig- Kumar, A., Srikanth, P., Nayyar, A., Sharma, G., Krishnamurthi, R., &
ital transformation of SMEs: A four levels approach. Management Deci- Alazab, M. (2020). A novel simulated-annealing based electric bus sys-
sion, 58(8), 1543–1562. https://doi.org/10.1108/MD-07-2019-0939 tem design, simulation, and analysis for Dehradun smart city. IEEE
Gioia, D. A., Corley, K. G., & Hamilton, A. L. (2013). Seeking qualitative Access, 8, 89395–89424. https://doi.org/10.1109/ACCESS.2020.
rigor in inductive research: Notes on the Gioia methodology. Organiza- 2990190
tional Research Methods, 16(1), 15–31. https://doi.org/10.1177/ Kumar, S., Sureka, R., Lim, W. M., Kumar Mangla, S., & Goyal, N. (2021).
1094428112452151 What do we know about business strategy and environmental
Glasser, B. G., & Strauss, A. L. (1967). The development of grounded theory. research? Insights from business strategy and the environment. Busi-
Alden. ness Strategy and the Environment, 30(8), 3454–3469. Portico. https://
Guba, E. G., & Lincoln, Y. S. (1981). Effective evaluation: Improving the use- doi.org/10.1002/bse.2813
fulness of evaluation results through responsive and naturalistic Li, X., & Maréchal, F. (2023). Deep excavation of the impact from endoge-
approaches. Jossey-Bass. nous and exogenous uncertainties on long-term energy planning.
Gupta, S., Modgil, S., Bhattacharyya, S., & Bose, I. (2021). Artificial intelli- Energy and AI, 11, 100219. https://doi.org/10.1016/j.egyai.2022.
gence for decision support systems in the field of operations research: 100219
Review and future scope of research. Annals of Operations Research, Li, Y., Li, S., Sun, X., & Hao, D. (2022). Prediction of carbon dioxide produc-
308(1–2), 215–274. https://doi.org/10.1007/s10479-020-03856-6 tion from green waste composting and identification of critical factors
Hai, T., Alsharif, S., Dhahad, H. A., Attia, E.-A., Shamseldin, M. A., & using machine learning algorithms. Bioresource Technology, 360,
Ahmed, A. N. (2022). The evolutionary artificial intelligence-based 127587. https://doi.org/10.1016/j.biortech.2022.127587
algorithm to find the minimum GHG emission via the integrated Lippiatt, N., Ling, T.-C., & Pan, S.-Y. (2020). Towards carbon-neutral con-
energy system using the MSW as fuel in a waste heat recovery plant. struction materials: Carbonation of cement-based materials and the
Sustainable Energy Technologies and Assessments, 53, 102531. https:// future perspective. Journal of Building Engineering, 28, 101062. https://
doi.org/10.1016/j.seta.2022.102531 doi.org/10.1016/j.jobe.2019.101062
Hannigan, T. R., Haans, R. F., Vakili, K., Tchalian, H., Glaser, V. L., Liu, Y., Esan, O. C., Pan, Z., & An, L. (2021). Machine learning for advanced
Wang, M. S., Kaplan, S., & Jennings, P. D. (2019). Topic modeling in energy materials. Energy and AI, 3, 100049. https://doi.org/10.1016/j.
management research: Rendering new theory from textual data. Acad- egyai.2021.100049
emy of Management Annals, 13(2), 586–632. https://doi.org/10.5465/ Liu, Z., Jiang, P., Wang, J., & Zhang, L. (2022). Ensemble system for short
annals.2017.0099 term carbon dioxide emissions forecasting based on multi-objective
Heo, S., Ko, J., Kim, S., Jeong, C., Hwangbo, S., & Yoo, C. (2022). Explain- tangent search algorithm. Journal of Environmental Management, 302,
able AI-driven net-zero carbon roadmap for petrochemical industry 113951. https://doi.org/10.1016/j.jenvman.2021.113951
 10990836, 2024, 5, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/bse.3689 by Bangladesh Hinari NPL, Wiley Online Library on [27/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
LUQMAN ET AL. 4003

Ma, H., Zhou, Z., Zhang, X., & Chen, X. (2023). Towards carbon-neutral occupancy heat emissions for demand-driven control solutions. Energy
edge computing: Greening edge ai by harnessing spot and future car- and Buildings, 226, 110386. https://doi.org/10.1016/j.enbuild.2020.
bon markets. IEEE Internet of Things Journal, 10, 16637–16649. 110386
https://doi.org/10.1109/JIOT.2023.3268339 Walsh, T., Evatt, A., & de Witt, C. S. (2020). Artificial intelligence & climate
Maxwell, S. E. (2000). Sample size and multiple regression analysis. Psychologi- change: Supplementary impact report. Oxford, 1, 1–15.
cal Methods, 5(4), 434–458. https://doi.org/10.1037/1082-989X.5.4.434 Wang, S., Li, J., Liu, X., Zhao, E., & Eghbalian, N. (2022). Multi-level charg-
Minevich, M. (2021). 11 examples of AI climate change solutions for zero ing stations for electric vehicles by considering ancillary generating
carbon. Available at: https://www.forbes.com/sites/markminevich/ and storage units. Energy, 247, 123401. https://doi.org/10.1016/j.
2021/10/08/11-examples-of-ai-climate-change-solutions-for-zero- energy.2022.123401
carbon/?sh=7806f2462251 WEF. (2022). The world economic forum. https://www.weforum.org/
Moktadir, A., Kumar, A., Ali, S. M., Paul, S. K., Sultana, R., & Rezaei, J. (2020). Whittemore, R., Chase, S. K., & Mandle, C. L. (2001). Validity in qualitative
Critical success factors for a circular economy: Implications for business research. Qualitative Health Research, 11, 117–132. https://doi.org/
strategy and the environment. Business Strategy and the Environment, 10.1177/104973201129119299
Portico, 29(8), 3611–3635. https://doi.org/10.1002/bse.2600 Wu, J., Wang, J., & Kong, X. (2022). Strategic bidding in a competitive elec-
Natgunanathan, I., Mak-Hau, V., Rajasegarar, S., & Anwar, A. (2023). Dea- tricity market: An intelligent method using multi-agent transfer learn-
kin microgrid digital twin and analysis of AI models for power genera- ing based on reinforcement learning. Energy, 256, 124657. https://doi.
tion prediction. Energy Conversion and Management: X, 18, 100370. org/10.1016/j.energy.2022.124657
https://doi.org/10.1016/j.ecmx.2023.100370 Xi, H., Wu, X., Chen, X., & Sha, P. (2021). Artificial intelligent based energy
Okuyama, A., Yoo, S., Kumagai, J., Keeley, A. R., & Managi, S. (2022). Ques- scheduling of steel mill gas utilization system towards carbon neutral-
tioning the sun: Unexpected emissions implications from residential ity. Applied Energy, 295, 117069. https://doi.org/10.1016/j.apenergy.
solar photovoltaic systems. Resources, Conservation and Recycling, 176, 2021.117069
105924. https://doi.org/10.1016/j.resconrec.2021.105924 Zhang, A., Alvi, M. F., Gong, Y., & Wang, J. X. (2022). Overcoming barriers
Petrescu, M., Ajjan, H., & Harrison, D. L. (2023). Man vs machine– to supply chain decarbonization: Case studies of first movers.
detecting deception in online reviews. Journal of Business Research, Resources, Conservation and Recycling, 186, 106536. https://doi.org/
154, 113346. https://doi.org/10.1016/j.jbusres.2022.113346 10.1016/j.resconrec.2022.106536
Pulselli, R. M., Marchi, M., Neri, E., Marchettini, N., & Bastianoni, S. (2019). Zhang, J., Huang, Y., Aslani, F., Ma, G., & Nener, B. (2020). A hybrid intelli-
Carbon accounting framework for decarbonisation of European city gent system for designing optimal proportions of recycled aggregate
neighbourhoods. Journal of Cleaner Production, 208, 850–868. https:// concrete. Journal of Cleaner Production, 273, 122922. https://doi.org/
doi.org/10.1016/j.jclepro.2018.10.102 10.1016/j.jclepro.2020.122922
Rocha, H. R., Honorato, I. H., Fiorotti, R., Celeste, W. C., Silvestre, L. J., & Zhao, J., Xi, X., Na, Q., Wang, S., Kadry, S. N., & Kumar, P. M. (2021a). The
Silva, J. A. (2021). An artificial intelligence based scheduling algorithm technological innovation of hybrid and plug-in electric vehicles for
for demand-side energy management in smart homes. Applied Energy, environment carbon pollution control. Environmental Impact Assess-
282, 116145. https://doi.org/10.1016/j.apenergy.2020.116145 ment Review, 86, 106506. https://doi.org/10.1016/j.eiar.2020.106506
Sadoudi, S., Boudour, M., & Kouba, N. E. Y. (2021). Multi-microgrid intelli- Zhao, L.-T., Miao, J., Qu, S., & Chen, X.-H. (2021b). A multi-factor inte-
gent load shedding for optimal power management and coordinated grated model for carbon price forecasting: Market interaction promot-
control with energy storage systems. International Journal of Energy ing carbon emission reduction. Science of the Total Environment, 796,
Research, 45(11), 15857–15878. https://doi.org/10.1002/er.6819 149110. https://doi.org/10.1016/j.scitotenv.2021.149110
Scherz, M., Hoxha, E., Kreiner, H., Passer, A., & Vafadarnikjoo, A. (2022). A Zhao, X., Ma, X., Chen, B., Shang, Y., & Song, M. (2022). Challenges toward
hierarchical reference-based know-why model for design support of carbon neutrality in China: Strategies and countermeasures. Resources,
sustainable building envelopes. Automation in Construction, 139, Conservation and Recycling, 176, 105959. https://doi.org/10.1016/j.
104276. https://doi.org/10.1016/j.autcon.2022.104276 resconrec.2021.105959
Shashi, M. (2022). Sustainable digitalization by leveraging digitainability Zhou, D., Hu, F., Zhu, Q., & Wang, Q. (2022a). Regional allocation of
matrix in supply chain. International Journal of Innovative Technology renewable energy quota in China under the policy of renewable port-
and Exploring Engineering, 11, 16–20. https://doi.org/10.35940/ijitee. folio standards. Resources, Conservation and Recycling, 176, 105904.
K9297.10111122 https://doi.org/10.1016/j.resconrec.2021.105904
Shi, T., Si, S., Chan, J., & Zhou, L. (2021). The carbon emission reduction Zhou, Q., Li, Y., Zhao, D., Li, J., Williams, H., Xu, H., & Yan, F. (2022b).
effect of technological innovation on the transportation industry and Transferable representation modelling for real-time energy manage-
its spatial heterogeneity: Evidence from China. Atmosphere, 12(9), ment of the plug-in hybrid vehicle based on k-fold fuzzy learning and
1169. https://doi.org/10.3390/atmos12091169 Gaussian process regression. Applied Energy, 305, 117853. https://doi.
Shukhobodskiy, A. A., Zaitcev, A., Pogarskaia, T., & Colantuono, G. (2021). org/10.1016/j.apenergy.2021.117853
RED WoLF hybrid storage system: Comparison of CO2 and price tar- Zhou, Y. (2022a). Artificial intelligence in renewable systems for transfor-
gets. Journal of Cleaner Production, 321, 128926. https://doi.org/10. mation towards intelligent buildings. Energy and AI, 10, 100182.
1016/j.jclepro.2021.128926 https://doi.org/10.1016/j.egyai.2022.100182
Su, Y., & Fan, Q.-M. (2019). The green vehicle routing problem from a Zhou, Y. (2022b). Ocean energy applications for coastal communities with
smart logistics perspective. IEEE Access, 8, 839–846. https://doi.org/ artificial intelligence—A state-of-the-art review. Energy and AI,
10.1109/ACCESS.2019.2961701 10, 100189. https://doi.org/10.1016/j.egyai.2022.100189
Talwar, S., Kaur, P., Okumus, B., Ahmed, U., & Dhir, A. (2021a). Food waste
reduction and taking away leftovers: Interplay of food-ordering routine,
planning routine, and motives. International Journal of Hospitality Man-
agement, 98, 103033. https://doi.org/10.1016/j.ijhm.2021.103033 How to cite this article: Luqman, A., Zhang, Q., Talwar, S.,
Talwar, S., Kaur, P., Yadav, R., Sharma, R., & Dhir, A. (2021b). Food waste Bhatia, M., & Dhir, A. (2024). Artificial intelligence and
and out-of-home-dining: Antecedents and consequents of the decision corporate carbon neutrality: A qualitative exploration. Business
to take away leftovers after dining at restaurants. Journal of Sustainable
Strategy and the Environment, 33(5), 3986–4003. https://doi.
Tourism, 31, 1–26. https://doi.org/10.1080/09669582.2021.1953512
Tien, P. W., Wei, S., Calautit, J. K., Darkwa, J., & Wood, C. (2020). A vision- org/10.1002/bse.3689
based deep learning approach for the detection and prediction of