Table 2.

Literature taxonomy of Artificial Intelligence implementation for supply chain sustainability

Author(s)	AI aim	AI implementation benefits	AI implementation challenges	Study nature	Study methodology
Baryannis et al. (2019)	Identification, assessment, mitigation and monitoring of SC risks	Enables accelerated and adaptive decision making based on large datasets Provides automated decision-making, predictive and learning capabilities Enables increased SC visibility	N.A.–N.S	Ql	Literature review
Camaréna (2020)	AI for transitioning food systems	Supports increase in productivity (production optimisation) Ensures automation of seeding, crop yield, harvesting, etc Informs decision-making Enables machine learning for detecting the decomposition rate of vegetables Enables machine learning and machine vision to detect parasites on fish Helps detect environmental pollution	Legal and ethical issues of responsibility and liability (e.g., through automated decision-making) Exacerbation of tensions between social and technical engineering Jeopardising of data confidentiality (e.g., by hacking by competitors and foreign powers) Privacy rights infringements of farmers by governments through applying AI for SC traceability Lack of required systems thinking by participants in the food system	Ql	Literature review
Chidepatil et al. (2020)	Blockchain and sensor-driven AI for transforming the circular economy	Facilitates information sharing among SC members Enables smart contracts in the procurement processes between SC members	Lack of reliable information about availability, quantity, quality, and suitability	Ql	Case study
Cubric (2020)	Drivers and barriers to AI adoption in business and management	Supports increased productivity Promotes cost reductions (e.g., reduced human errors, equipment cost, reduction of human labour, etc.) Supports decision-making (e.g., informed decisions, more accurate forecasting, etc.) Outlines more sustainable processes to fit consumer demand (e.g., agriculture)	Labelling/structuring data can be costly Support of infrastructure is necessary Lack of training data may lead to reduced performance of AI algorithms Data may be unstructured and thus challenging to share Data confidentiality issues Generalisability of data may not be possible Difficulties in reusing AI models for different purposes Lack of knowledge about the benefits of AI for specific problems	Ql	Literature review
Dauvergne (2020)	Environmental impact of AI	Promotes corporate efficiency and productivity gains Enhances energy efficiency of transnational corporations, e.g., automation of cooling systems with AI Promotes efficiency and reliability gains of renewable energies, e.g., forecasting weather, fine-tuning energy storage, etc Helps reduce fuel consumption by increasing efficiency in production and logistics Supports programs to cut costs of packaging and shipping	Negative environmental impact Increases in consumption and production outweigh positive effects on corporate social responsibility Enhanced social inequality Accelerated natural resources extraction Racial and ethnic profiling of customers	Ql	Literature review
Di Vaio et al. (2020)	Impact of AI on agri-food SCs	Promotes testing concerning food safety at every stage of the SC Promotes increases in efficiency and productivity Enables food savings, improves the hygiene of production sites, and helps clean up production equipment quickly Enables food sorting (e.g., optical sensor-based solutions with machine learning capabilities) Helps ensure hygiene standards (e.g., use of cameras to monitor the compliance with hygiene standards) Helps the use of drones to assess the ripening status of crops	Enhanced social inequality Privacy issues due to AI application data Required significant changes in business models	Ql	Literature review
Ebinger and Omondi (2020)	Transparency in sustainable SCs	Enables traceability and tracking through the collection and processing of big data for end-to-end SC visibility Fosters cooperation and partner selection via real-time information sharing for decision-making processes Promotes governance via end-to-end SC data collection and processing Assists in strategic and operational risk assessment via data analytics	Individualisation of digital solutions in companies via using internal data architecture Lack of standardised information leading to difficulties in choosing appropriate solutions to increase transparency Many approaches used are still in the trial phase and offer only limited solutions	Ql	Literature review
Min (2010)	Support of decision-making: strategic/ operational/ tactical	Provides a decision support tool that can help firms connect with customers, suppliers and SC partners Facilitates information exchange among SC entities Facilitates understanding of SC dynamics Supports the logistics outsourcing or contract manufacturing decisions Supports supplier selection Facilitates real-time pricing and reverse auctioning involving SC partners	Solutions may be too costly Difficult to produce solutions due to the complexity and ill-structured nature of the problem Relative youth and broad spectrum of the SC management discipline AI solutions may be too difficult for decision-makers to comprehend Difficulties in implementation for handling risks in cross-functional or cross-border contexts due to knowledge acquisition bottlenecks	Ql	Literature review
Mota et al. (2015)	AI for economic, environmental and social design and planning	Helps improve all sustainability dimensions (i.e., economic, environmental, social) Promotes cost reductions and mitigation of environmental hazards through modelling (e.g., reducing warehouses) Helps increase social sustainability with a slight compromise of economic performance	Indicators for sustainability in the dimensions considered for this work are imperfect/can be significantly improved No possibility to evaluate the social dimension of an SC per se Consideration of only a single SC	Qn	Generic multi-objective mathematical programming model
Orji and Wei (2015)	Fuzzy logic and system dynamics for sustainable supplier selection	Helps in the modelling of supplier selection criteria for decision support	No spatial and temporal considerations regarding decisions Criteria and alternatives in the model are fixed	Qn	Fuzzy logic and system dynamics
Sanders et al. (2019)	AI for sustainable SCs	Helps improve demand management Promotes increased operational productivity Promotes increased operational efficiency Enables increased SC transparency, traceability and security Supports decision support (e.g., pricing)	Infliction of severe economic and social cost Creation of biases and discrimination through machine learning applications Potential failure of AI algorithms to detect misinformation	Ql	Literature review
Ting et al. (2014)	Quality assurance in food SCs	Supports decision-making Helps create and transfer quality relevant information Promotes inspection cost reductions Promotes quality level enhancements Promotes increased customer satisfaction Enhances SC visibility	Increasing costs for transportation Requirement of significant time to identify dominant association rules by computational methods	Qn	Association mining

AI Artificial intelligence; SC Supply chain; Ql. Qualitative; Qn. Quantitative