Artificial Intelligence in Supply Chain Management Compliance

by Kanerva Jalas

Artificial Intelligence (AI) has caused large-scale discussion in relation to human rights and the labour market; the consequences that the technology might have for the future of humans has caused a variety of reactions: fear of AI overtaking jobs, resentment towards digitalisation, and a demand for regulation. Amidst these discussions, it can be helpful to realise for what purpose AI serves in most contexts: a useful tool which can be employed for time-consuming and tedious tasks. Such an instance is global supply chain management. The term ‘supply chain’ refers to the process of manufacturing a good and providing it to the market for consumers. A global supply chain on the other hand is the same evolution of a product but the chain of steps in this process takes place in several countries. This development of international trade has resulted in a problematic consequence: firms often headquartered in developed nations will benefit from labour in countries with weaker governance, resulting in exploitation of, for example, considerably lower wages, unsafe working conditions and poor employment safety regulation. These circumstances have been seen occasionally to result in tragic outcomes: the collapse of Rana Plaza factory in 2013 caused the deaths of over 1100 workers.

To prevent such accidents, regulations have been created on an international as well as national level: the United Nations Guiding Principles on Business and Human Rights (2011) includes three pillar levels including 1) a state duty to protect, 2) a corporate responsibility and 3) requirement of access to remedies. The UN principles, however, are not binding international law. For more effective enforcement, some countries have enacted national legislation. In 2015, the United Kingdom introduced the Modern Slavery Act requiring supply chain disclosure. Similar legislative instruments have been passed, for example, in the United States, Norway and Australia. These laws require undertakings to disclose relevant information about their supply chains, such as due diligence actions. In practice, the requirements imposed on companies can be seen as transparency reports on websites or physical locations.

Despite the growing efforts to fight human rights issues arising from globalisation of business, disclosure and transparency legislation has done so to a limited extent. A lack of oversight exists as to which companies are reporting, and fundamentally, the quality of the disclosure reports. Alternatively, where such an effort exists, it is time-consuming and difficult work: the most large-scale study up-to-date included investigations of just 900 reports, and took over two years to complete. Therefore, it is unlikely that the thousands of reports yearly are verified to be in compliance.

AI can be defined as algorithms that have the ability to gather, analyse and interpret data, and to use that data to achieve its learning goals. Therefore, compliance with global supply chain regulations is a task proposed suitable for AI. However, a multitude of challenges are faced with this approach. First, despite the fact that many national legislation imposes companies to specify their practices in ensuring legality of their global supply chain systems, there are no requirements on how specific these reports should be. The lack of a standardised system imposes difficulties for AI analysis. Second, an AI system does not have adequate access to the reports. In order to acquire the relevant knowledge regarding company practises, a system would have to exist which mandates identification of those undertakings falling under the scope of disclosure regimes. One study regarding the accessibility of disclosure reports found that around only 12,000 of a total 17,000 reports were available. Within these 12,000, almost a 5000 had access errors related to their HTML format. A similar problem was found to be present with reports that are offered in PDF format. The introduction of such technology for the field has been seen in practice to an extent: in a joint cooperation with IBM, the delivery company DHL introduced ‘supply chain applications’ to boost efficiency of transportation. Such applications included, for example, the AI scheduling pick-up appointments, and maintenance in accordance with weather circumstances and traffic. Similarly, some companies have benefited from the use of machine learning in the context of social media and news, to locate supply chain interferences.

Despite the challenges, the introduction of an AI system for global supply chain management is promising, but requires further implementation. First, it must be imposed on undertakings with supply chains to publish disclosure reports according to a standards format. This would allow for machine-readable collection of data which would allow for adequate comparison and assessment of legality of measures.

Sources:

D Nersessian and D Pachamanova, ‘Human Trafficking in the Global Supply Chain: Using Machine Learning to Understand Corporate Disclosures under the UK Modern Slavery Act’ (2022) 35 Harvard Human Rights Journal 1.

N Weinberg and others, ‘AI Against Modern Slavery: Digital Insights into Modern Slavery Reporting’ (2020) AAAI Symposium on AI for Social Good.

A Behladi and others, ‘Artificial Intelligence-Driven Innovation for Enhancing Supply Chain Resilience and Performance Under the Effect of Supply Chain Dynamism: An Empirical Investigation’ (2021) Annals of Operations Research.

S Chilton and A Sarfaty, ‘The Limitations of Supply Chain Disclosure Regimes’ (2017) 53 Stanford Journal of International Law 1.

P Tambe and P Tambay, ‘Reducing Modern Slavery Using AI and Blockchain’ (2020) International Conference on Artificial Intelligence for Good.

Computing Community Consortium, ‘Applying AI in the Fight Against Modern Slavery’ (2021): https://arxiv.org/ftp/arxiv/papers/2106/2106.13186.pdf, accessed on the 16th February 2023.