1. Introduction

2. Literature Review

3. Methodology

4. Results

5. Analysis & Discussion

5.1. AI—Technology Characteristics

Insights concerning AI’s capability that resulted from the study of the interviews include its applicability for identifying and monitoring diverse events in a vast array of domains. According to the findings section and the literature section (Dhall et al., 2020), when an event is well understood,

AI may be taught to recognise the patterns of these events and then search for and identify them in everyday real-world data if the patterns are well-understood, quantitative, and data-rich. Several application areas, such as detecting and monitoring deforestation or the expansion of spruce beetles and recognising particular tree species groups, were highlighted in the empirical findings and the literature (Ritter et al., 2017). As shown by empirical evidence, AI generates novel ideas that would be too difficult or costly for humans to produce. Moreover, as shown by the findings and literature (Akerkar, 2018; Dhall et al., 2020), with extra time and data, monitoring may be replaced by predictive capabilities, where AI begins to recognise the beginning patterns of an event. These include, as shown by the empirical data, the monitoring and forecasting of three growth rates and the prediction of deforestation. Similarly, a survey of the relevant literature reveals that AI is being used for more basic predictive functions, such as demand forecasting (Cubric, 2020; Agrawal et al., 2017), as well as more complex instances, such as rainfall prediction (Pham et al., 2020).

Figure 1. Task-technology fit framework.

Figure 1. Task-technology fit framework.

AI’s ability to enhance scenario planning by monitoring and anticipating real-time happenings is another capability applicable to sustainable sourcing. As described in the literature (Dhall et al., 2020; Agrawal et al., 2017; Cubric, 2020; Pham, 2020; Burgess, 2018) and shown by the empirical results, AI may be taught to predict distinct occurrences and their probabilities using accessible data.

An intriguing conclusion from the empirical data cited by several responders is that, because of these characteristics, AI has the ability to aid in the modelling of many situations, as validated by the literature (Baryannis et al., 2019). Consequently, although humans must choose which scenarios to follow and which risks to take, AI may provide people with the potential consequences of their actions (Agrawal et al., 2017; Wilson and Daughtery, 2018). Moreover, according to one respondent, a growth in the number of accessible smart devices increases the overall amount of data and permits the integration of these characteristics into new domains.

Literature (Herweijer & Waughray, 2018) and empirical findings demonstrate that AI can automate a number of manual operations that would otherwise need human interaction. In addition, by training algorithms using AI techniques such as ML to accomplish certain jobs, AI may drastically lower the costs associated with this sort of labour (Agrawal & Kirkland, 2018). Moreover, as demonstrated by the literature (Herweijer & Waughray, 2018; Agrawal & Kirkland, 2018) and empirical findings, AI can deliver these results faster, more accurately, and at a lower cost than humans can, owing to its scalability and ability to continuously process large amounts of data, which boosts the overall productivity of a business.

Figure 2. Machine Learning Process.

Figure 2. Machine Learning Process.

5.2. Applied Framework

The empirical data and literature study highlighted three sustainable sourcing challenges: supply chain difficulties, information overload, and effect evaluation and projections. In the following sections, AI traits and the task will be assessed.

IKEA interviewees highlighted supply chain concerns as a major issue. As several IKEA respondents stated, due to the inherent complexity of current supply chains caused by their increasingly globalised and diverse supply chains, with countless sub-suppliers and often without direct contact with suppliers at the raw materials’ origin, IKEA and its suppliers must spend a lot of money and time to understand where their raw materials are coming from. Many IKEA respondents say this complexity hinders IKEA’s traceability and transparency. Literature also links multi-tier supply chains to poor transparency (Mejas, Bellas, Pardo & Paz, 2019). IKEA respondents acknowledge that many enterprises have trouble mapping their supply networks, partly owing to their tendency to employ complicated supply chains and partly due to sophisticated traceability requirements in certain areas, notably the EU’s lumber regulation (Pajares et al., 2020). Some IKEA responders say this problem doesn’t affect all goods. Wood planks have simpler supply chains and are easier to monitor, offering more traceability and transparency. Thus, supply chain complexity and opacity, which rely on product raw materials, affect traceability and transparency.

Businesses must account for the costs and benefits of their operations for traceability and resource allocation, which makes resource balancing tough. Some respondents said that sustainability employees struggle to determine whether information is useful. Empirical research suggests resource balancing concerns when legislative mandates that may have little beneficial effects on sustainability incur huge expenses that might have been employed on enterprises that greatly influence sustainability.

One IKEA responder noted that suppliers typically focus on compliance and miss out on environmental activities.

The responder further noted that providers may be tempted to fabricate papers due to these time-consuming traceability requirements. In conclusion, traceability is essential for sustainable supply chain practises (Van den Brink et al., 2019), but as shown by empirical data, the demands traceability places on companies, especially for harder-to-trace raw materials, can lead to disproportionate resource allocation compared to their sustainability impact. As indicated in the empirical data, guaranteeing sustainability from the local source to the focus company is vital since most environmental harm occurs there, a conclusion backed by literature (Sanders et al., 2019). Thus, measuring and mitigating sustainability consequences necessitates long-term supply chains all the way to the final provider.

According to several respondents, organisations that IKEA does not have direct legal connections with are tougher to target for traceability and sustainability projects since IKEA cannot place as much pressure on them as on its direct suppliers, which makes traceability difficult and time-consuming. Wilhelm, Blome, Wieck, and Xiao (2016) note that many organisations lack direct legal contracts with suppliers in multi-tier supply chains, making it difficult to implement sustainability programmes. As noted in the empirical chapter, IKEA typically relies on supplier documentation that travelled across levels or had to be obtained by contacting suppliers on other levels, which raises the potential for mistakes or intentional misbehaviour. Some IKEA respondents noted that entering data into origin tracking systems manually is time-consuming and error-prone. A previous study acknowledged this issue (Leong et al., 2018).

Verifying the tracing documentation is another empirical hurdle. One IKEA responder said suppliers have varied internal information reporting systems and processes, sometimes even relying on paper documentation, which increases verification documentation collection time and quality. The empirical findings also suggest that some suppliers are reluctant to disclose all the essential information since it puts them at risk of being bypassed or undercut. This limits IKEA’s traceability and transparency. The literature also notes that enterprises struggle to acquire data from their suppliers, which makes it difficult for them to create sustainable supply chains (Gurtu et al., 2017). IKEA’s sustainable sourcing approach is hampered by verification and information sharing.

AI has many traits that match the first sustainable sourcing difficulty: supply chain difficulties. As mentioned above, IKEA struggles to gather information, which affects traceability and transparency and leads to a lack of useful data. As some AI respondents noted, using AI’s ability to monitor certain aspects and developments at and around suppliers’ operations could create the possibility of monitoring development on the ground, such as deforestation monitoring, mining operations, etc., allowing firms like IKEA to access previously difficult-to-obtain information and even verify supplier information. This might assist organisations in detecting and addressing misbehaviour, minimising the risk of non-compliance, and increasing sustainability in the real world. Other AI solutions that gather end-to-end data might help ease traceability by collecting data along the supply chain (Ebinger & Omondi, 2020). While this solution does not directly address the challenge of gaining access to some information that certain suppliers are unwilling to share and the problem with falsified documents remains, it does circumvent those challenges by delivering some of the needed information without the possibility of tampering and giving IKEA real-world insights.

Many AI responders said alternative technologies can help firms with these difficulties.

AI benefits from its scalability (Agrawal & Kirkland, 2018) and empirical outcomes. AI’s capability can be doubled, quadrupled, etc., but expenses rise just a little (Agrawal & Kirkland, 2018). This makes it possible to continually monitor full supply chains instead of inspecting a limited sample of vendors, boosting the possibility of discovering wrongdoing when data is available. AI’s neutrality may also help organisations since it has fewer prejudices than people, reducing human error (Akerer, 2018). Thus, AI can objectively check more sites than humans. These extra AI advantages may generate a clear opportunity for AI in sustainable sourcing, potentially benefiting organisations. Some AI responders said monitoring might become prediction, allowing companies to act before certain events. These predictive capabilities might be linked into strategic scenario planning systems to help managers make the optimal decisions (Herweijer & Waughray, 2018; Agrawal et al., 2017; Cubric, 2020; Pham, 2020).

The actual data showed that sub-supplier tracking was laborious. Suppliers often use different systems, so they have to manually transfer data from their internal systems into IKEA’s tracing systems, sometimes from unstructured data like handwritten paper sheets, which increases the chance of human error and makes it laborious, reducing the likelihood of getting all the needed data in good quality. As one AI responder noted, AI may free up resources for sustainability or value-added initiatives. AI-trained algorithms might harvest data instead of replicating it, automating certain monotonous, ongoing procedures (Akerer, 2018). This suggests that AI might help suppliers improve, especially if they already have data in their internal systems that they manually upload to IKEA’s tracing systems. AI-automated methods might save time and money (Herweijer & Waughray, 2018; Agrawal & Kirkland, 2019).

Human mistakes may decrease when manual input decreases, enhancing data reliability. Supplier data limits all of this. This strategy would not work if suppliers used paper-based bookkeeping instead of digitised information systems, but other AI approaches that concentrated on unstructured data may (Akerkar, 2018). As shown in the empirical section, only a percentage of the required information can be gathered this way; hence, AI’s ability to help enterprises collect all the essential information depends on their suppliers’ and sub-suppliers’ digitalization. Centralized systems may also solve this problem more easily.

According to multiple replies, many organisations, including IKEA, are moving their supply chains from compliance to a more holistic development strategy that positively influences the actual world. In the empirical portion, corporate ethics, children’s rights, food security, land rights, biodiversity, land-use, traceability, and transparency are evaluated concurrently and in connection to each other when examining raw material extraction and production. Some IKEA respondents said this requires more data to be gathered, processed, and analysed to guarantee that actions have meaningful and positive implications for sustainability. A more comprehensive strategy requires more data and the capacity to acquire and interpret it, as Notar-Nicola et al. (2017) and Ritter (2017) noted. As shown in the empirical section, identifying every sustainability factor is complicated, which complicates the holistic approach, as supported by the research (Gurtu et al., 2017).

Another intriguing result from the data was that certain IKEA respondents felt a need for automated, rapid, and real-time analysis that can handle several variables and parameters, assess and analyse this information, and allow action in real time. However, as one IKEA responder noted, the information frequently gets too complicated, hindering action where it is required. One respondent said that the supply chain’s structure, related to the preceding chapter on supply chain concerns, became exceedingly complicated and intertwined, making it impossible to analyse and comprehend. According to Ebinger and Omondi (2020), this creates vast volumes of data that companies must process, which may increase human mistakes.

Another probable match was found. AI’s speed and pattern recognition make it handy here. (2018) AI-trained algorithms can accurately recognise multi-dimensional patterns that would normally require experienced people. 2018. IKEA’s holistic approach requires more data to be assessed and analysed, and AI’s capacity to recognise patterns in vast volumes of data and provide unique insights (Akerkar, 2018) should aid enterprises with the new dataflows, according to empirical studies. As shown in the empirical section, AI may help organisations analyse data and provide new insights that are readily understood and actionable. AI can also process data instantly (Ebinger & Omondi, 2020). Some respondents claim AI cannot manage, rank, or act on these findings. As one commenter noted, AI systems may generate insights, but acting on them typically needs human engagement. In restricted AI applications, which need human interaction, AI cannot act on insights outside of its programmed scope.

IKEA respondents complain about the absence of effect evaluation and projection. As mentioned in the empirical part, owing to the complexity of certain sustainability issues, there are no uniform definitions or clear measurements, making real-world effect evaluations problematic. As demonstrated in the empirical findings, these impact evaluations often score company activities rather than assess their real-world implications. Several scholars (Lenzena et al., 2006; Ritter et al., 2017) have noted that comprehensive analyses of enterprises’ environmental impacts are rare and frequently neglect key aspects. The absence of uniform criteria and metrics makes effective forecasting and evaluation difficult for businesses (Gurtu, Searcy, and Jaber, 2017).

As other respondents noted, some organisations are inventing their own definitions of concepts and impact methodology, which might hinder sustainability since they can build their own individualised ways that may not be in accordance with a scientific, institutional understanding of sustainability.

Some respondents said it’s hard to understand how complicated anthropogenic concerns like climate change would affect organisations’ sourcing. One respondent underlined that to cope with these complex difficulties, it is necessary to comprehend the future consequences of numerous sustainability-related concerns, such as climate change, to guarantee long-term, profitable, sustainable raw material procurement and management.

This section discovered a possible challenge. Some responses suggested using AI for impact and prediction. (Dhall et al., 2020) AI may be taught to recognise patterns, search for them in daily data, and anticipate them by recognising early signals of pattern recurrence. As shown in the empirical section, AI may be used to monitor deforestation and spruce beetle spread using satellite images, surveillance films, tracking devices, etc. This allows IKEA to continuously monitor the environmental implications of its supply chains and how their sustainability efforts affect them. As before, scenario modelling and predicted effect monitoring should be improved.

AI and big data might analyse the implications of climate change. The lack of standard definitions and measurements, as noted by IKEA and AI respondents, limits the ability to analyse benefits using AI or traditionally.IKEA responders showed that biodiversity definitions and measurements are still being developed. Without them, assessing outcomes’ effects is difficult and inadequate, limiting AI’s present use. Several respondents noted that data for these evaluations is a possible restriction, and the research emphasises data as a key component of AI (Cubric, 2020). Thus, AI can only help with effect assessment and prediction in a small percentage of situations, which may be beneficial for certain organisations if those cases are worthwhile.

Some respondents claimed that AI can forecast future recurrent situations when difficulties are clearly defined, causes are recognised, and data is accessible and of excellent quality. AI, like other forecasting methods, cannot foresee non-regular events like the COVID-19 pandemic. Empirical findings reveal that humans still struggle to grasp events like weather forecasts; therefore, utilising AI to anticipate their outcomes would likely provide similar results. As shown in the findings section, modern AI cannot address issues that even humans cannot grasp, but it may be an asset when causation, metrics, and historical data are available.

Figure 3. Summary of the theoretical findings.

Figure 3. Summary of the theoretical findings.

5.3. Overall Comments on AI in Sustainable Sourcing

Overall, both the empirical data and the literature demonstrate the absence of ready-to-use generic technical solutions adopted by a wide variety of players. In contrast, the majority of accessible and proposed solutions were case-specific, and many of the AI responders mentioned additional technologies or solutions that may better meet the demands, therefore assisting businesses in overcoming the sustainable sourcing difficulties. Consequently, the substance behind the hoopla surrounding current AI may be questioned. Yet, AI has enormous promise when the qualities of the problem at hand and AI’s traits align, as shown in earlier sections. Additionally, as acknowledged in the literature (Brynjolfsson Burgess, 2018) and the empirical section, there are new potential for AI in several domains, including sustainable sourcing, as a result of an increase in data and technical advancements.

Figure 4. Inclusion and exclusion criteria.

Figure 4. Inclusion and exclusion criteria.

6. Conclusions

References

1. Accenture. (2020). What is Artificial Intelligence?. Retrieved from [2021-05-16]:.
2. Agrawal, A., Gans, Joshua S, & Goldfarb, A. (2017). What to expect from artificial intelligence. MIT Sloan Management Review, 58(3), 23.
3. Agrawal, V., & Lee, D. (2019). The Effect of Sourcing Policies on Suppliers’ Sustainable Practices. Production and Operations Management, 28(4), 767-787. [CrossRef]
4. Agrawal, A., & Kirkland, R. (2018, May 11). The economics of artificial intelligence. McKinsey & Company. Retrieved from [2021-05-16]: https://www.mckinsey.com/business-func-tions/mckinsey-analytics/our-insights/the-economics-of-artificial-intelligence.
5. Ahi, P. and Cory S. (2013). A Comparative Literature Analysis of Definitions for Green and Sustainable Supply Chain Management. Journal of Cleaner Production, 52 (2013): 329-41. [CrossRef]
6. Akerkar, R. (2018). Artificial Intelligence for Business (Springer Briefs in business). Cham:Springer International Publishing AG. [CrossRef]
7. Alicke, K., Dilda, V., Görner, S., Mori, L., Rebuffel, P., Reiter, S. & Samek, R. (2020). Retrieved from [2021-05-16]: https://www.mckinsey.com/industries/metals-and-mining/our-in-sights/succeeding-in-the-ai-supply-chain-revolution.
8. Baryannis, G., Validi, S., Dani, S., & Antoniou, G. (2019). Supply chain risk management and artificial intelligence: State of the art and future research directions. International Journal of Production Research, 57(7), 2179-2202. [CrossRef]
9. Bearman, M., Smith, C.D., Carbone, A., Slade, S., Baik, C., Hughes-Warrington, M. & Neumann, D.L. (2012) Systematic review methodology in higher education, Higher Education Research & Development, 31:5, 625-640. [CrossRef]
10. Bell, E. Bryman, A. & Harley, B. (2019). Business Research Methods. 5th edition., Oxford: Oxford University Press.
11. Bergstein, B. (2019). From intelligent systems to intelligent organizations. Research-Technology Management, 62(3), 31-37. [CrossRef]
12. Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77-101. [CrossRef]
13. Brown, J.R., Bushuev, M.A., Kretinin, A.A., & Guiffrida, A.L. (2017). Recent Developments in Green Supply Chain Management: Sourcing and Logistics. In Khan, M., Hussain, M., & Ajmal,M.M. (Ed.), In Green Supply Chain Management for Sustainable Business Practice (pp. 191-217). IGI Global. [CrossRef]
14. Manyika, J., Lund, S., Chui, M., Bughin, J., Woetzel, J., Batra, P., & Sanghvi, S. (2017). Jobs lost, jobs gained: Workforce transitions in a time of automation. McKinsey Global Institute,150. Retrieved from [2021-05-16]: https://mck.co/3vX2eEh.
15. Burgess A. (2018) AI in Action. In: The Executive Guide to Artificial Intelligence. Palgrave Macmillan, Cham. [CrossRef]
16. Canhoto, A., & Clear, F. (2020). Artificial intelligence and machine learning as business tools: A framework for diagnosing value destruction potential. Business Horizons, 63(2), 183-193. [CrossRef]
17. Carter, C.R. and Liane Easton, P. (2011), “Sustainable supply chain management: evolution and future directions”, International Journal of Physical Distribution & Logistics Management, Vol. 41 No. 1, pp. 46-62. [CrossRef]
18. Chase, R.B., & Jacobs, F.R. (2018). Operations and Supply Chain Management, 15e (SIE).McGraw-Hill Education.
19. Chui, M., & Malhotra, S. (2018, November 13). AI adoption advances, but foundational barriers remain. McKinsey & Company. Retrieved from [2021-05-16]: https://www.mckinsey.com/featured-insights/artificial-intelligence/ai-adoption-advances-but-foundational-barriers-remain#.
20. Chui, M. (2017). Artificial intelligence the next digital frontier. McKinsey and Company Global Institute, 47(3.6). Retrieved from [2021-05-16]: https://mck.co/3vXWlXk.
21. Copeland, B.J. “Artificial intelligence”. Encyclopedia Britannica, 11 Aug. 2020, Retrieved from [2021-05-16]: https://www.britannica.com/technology/artificial-intelligence.
22. Cubric, Marija. (2020). Drivers, barriers and social considerations for AI adoption in businessand management: A tertiary study. Technology in Society, 62, 101257. [CrossRef]
23. Dhall D., Kaur R., Juneja M. (2020) Machine Learning: A Review of the Algorithms and Its Applications. In: Singh P., Kar A., Singh Y., Kolekar M., Tanwar S. (eds) Proceedings of ICRIC 2019. Lecture Notes in Electrical Engineering, vol 597. Springer, Cham. https://doi.org/10.1007/978-3-030-29407-6_5. [CrossRef]
24. Dauvergne, P. (2021). The globalization of artificial intelligence: consequences for the politics of environmentalism. Globalizations, 18(2), 285-299. [CrossRef]
25. Delua, J. (2021). Supervised vs. Unsupervised Learning: What’s the Difference? Retrieved from [2021-05-16]: https://www.ibm.com/cloud/blog/supervised-vs-unsupervised-learning.
26. Ebinger, F., & Omondi, B. (2020). Leveraging Digital Approaches for Transparency in Sustainable Supply Chains: A Conceptual Paper. Sustainability (Basel, Switzerland), 12(15), 6129. [CrossRef]
27. Elliot, B. & Andrews, W. (2017). A Framework for Applying AI in the Enterprise. Retrieved from [2021-05-15]: https://emtemp.gcom.cloud/ngw/globalassets/en/doc/documents/3751363-a-framework-for-applying-ai-in-the-enterprise.pdf.
28. Encarnación Moral-Pajares, Concepción Martínez-Alcalá, Leticia Gallego-Valero, & Ángela Andrea Caviedes-Conde. (2020). Transparency Index of the Supplying Countries’ Institutions and Tree Cover Loss: Determining Factors of EU Timber Imports? Forests, 11(1009), 1009. [CrossRef]