How might Artificial Intelligence transform corporate sustainability policies?

20 March, 2018

Rapid technological development in the field of Artificial Intelligence (AI) has triggered a growing debate about its ethical, political and legal implications for our daily lives. In this post, I suggest that one important area where these debates will play out is in the sustainability policies of transnational corporations (TNCs) towards their global supply chains.

Efforts to develop AI and blockchain to improve supply chain traceability are still pretty new and the focus so far has mainly been limited simply to ensuring products can be traced from lower tiers of the supply chain to supermarket shelves. There are even fewer initiatives that focus on sustainability, and these have largely been directed at environmental sustainability; so far little attention has been paid to the potential of AI to address labour and human rights issues.

However it is clear that AI will have a real impact in this area and that it will directly affect the existing relationships between TNCs, their suppliers, workers and customers. Think about how brands such as H&M and M&S source their products. In numerous countries around the globe, millions of workers in literally thousands of workplaces are busy making products for these brands. Each year, a mountain of auditing reports are published detailing working conditions within supplier firms in numerous countries, each with its own distinct local structure of industrial relations. Within this context, as many corporations – including retailers – begin to benefit from AI to attract customers and manage their purchases, sustainability departments can expect significant disruption from these technological shifts, to which they will have to adapt.

Technology and the transformation of global supply chain management

Therefore, at the dawn of a new technological revolution that is set to transform global supply chain management, it’s high time we launch a debate about the role of AI in industrial relations. This is not just in order to stay true to the creed, 'ethics by design', but also as a requirement of innovation. Debate will help us evaluate and develop existing approaches – such as human rights due diligence, as promoted by the UN Guiding Principles on Business and Human Rights – so we can continue to make progress in this area. In this way, all social partners – including employers, workers and other related entities such as NGOs or state authorities – will have greater scope to collaborate in order to assess current and potential risks to employees within the workplace and take steps to mitigate them.

Thus AI can analyse a vast amount of data very quickly and offer summary judgements that can be used to inform decision making. However, a primary concern in this context is the extent to which AI analysis can be relied upon to produce objective judgments that do not simply reproduce and legitimise existing discrimination or inequalities. This concern has prompted discussions around the accountability and transparency of algorithms and there have been efforts to understand how to mitigate the potentially discriminatory and unfair decisions of algorithms in a range of important areas of life, such as, say, applying for a bank loan, or seeking justice. Such ethical concerns are salient in industrial relations between social actors who are located within the hierarchical structure of the global supply chain of TNCs. The human rights due diligence approach has evolved to answer concerns over the impact of this hierarchical structure after many years of conflict between TNCs, suppliers and workers/trade unions.

For instance, if AI begins to play a role in the decisions made by TNCs about their suppliers (which supplier should get what order volume), then we need to answer the question: 'How can suppliers and workers/trade unions participate in this process?' In other words, how can they meaningfully influence the machine learning process? Recent efforts to develop machine learning systems that can provide a human-understandable rationale for their decisions might be of benefit in this context.

This discussion shows that a central issue will be how to teach machines about human and labour rights. Although TNCs are profit-driven entities, long years of struggle have taught all social actors that ethical principles are essential for sustainable business. So, certain thresholds of principles and rules should be raised to balance the competing interests of TNCs, suppliers and workers and other social partners so that all can have access to the teaching of machines.

AI promotes human rights

If AI internalises accumulated knowledge about 'business and human rights' and allows other social actors to track its decision-making process, then AI could even promote the due diligence approach.

For instance, suppliers frequently complain about contradictions between the demands of the purchasing and sustainability departments of TNCs. While the former try to lower prices and force suppliers to deliver goods as quickly as possible, sustainability departments ask suppliers to invest in promoting working conditions and recognise the fundamental rights of workers, including the rights to organise and bargain collectively. AI could provide better coordination better various departments and, based on the pre-determined principles and rules, could reward local suppliers that invest in sustainability as well.

Another example concerns the thousands of suppliers from many countries that compete with each other to attract orders. They accept audit missions from the various departments of the TNCs with which they trade. One department audits quality, another checks sustainability, yet another bargains over prices... Thousands of reports from all over the world fly off to the TNCs’ headquarters or regional offices, and each supplier is analysed by more than one report. This can create two major problems if:

  • these reports are not be evaluated objectively at HQ, and
  • local auditors do not have sufficient knowledge about the objective conditions of workplaces and industrial relations in a given region or country.

AI could evaluate the vast amount of data that arises from these processes objectively. It could harmonise auditing practices by taking account of local realities and assess the historical account of suppliers and determine current and potential risks that demand a response. Such an objective approach would also make it easier to resolve potential disputes

Sustainability departments as champions of ethical AI

The working methods of sustainability departments and their basic principles are standard for many TNCs and companies tend to collaborate on sustainability via various multi-stakeholder initiatives (MSIs) that also include trade unions, NGOs and local suppliers. This is a clear space in which sustainability departments can become pioneers of ethical and accountable AI.

Even as 'commercial confidentiality' serves as a blanket justification for companies to refuse to disclose their algorithms, the fact that all major corporations agree on certain ethical principles means they should cooperate under MSIs to develop and apply AI for sustainable and ethical trade. MSIs may also recognise scientists, engineers and their associations/institutes involved with AI as their new stakeholders.

For instance, MSIs such as the Ethical Trading Initiative or the Fair Wear Foundation should act as watchdogs for the AI practices of their member corporations and should allow suppliers and trade unions to access the machine learning processes involved in developing them. Such a process will open a door for trade unions to get involved in the design process in addition to discussing the future of the work.

Such a collaboration among corporations, trade unions and NGOs and joint investment would pave a new way for ethical AI to feed into more ethical supply chain management.

Subscribe to the blog

About the author(s)
Emre Eren Korkmaz
Departmental Lecturer in Migration and Development