Human rights, a term that can evoke a range of reactions. In the context of this article, human rights are defined as the basic rights every human being is entitled to enjoy regardless of their place of birth, beliefs, sexual orientation, or gender amongst other characteristics or attributes. So, how could we achieve that?
We begin with commonalities. Every human, for example, fundamentally wants the opportunity to earn enough income to put sufficient food on the table, a roof over their head(s), provide for basic needs and do so in a safe environment. When we agree on these fundamental concepts, we are able to agree on the need to ensure these basic rights are protected for all.
Setting the Stage
The International Labour Organization (ILO), International Organization for Migration (IOM) and international human rights group Walk Free published a report (2022) revealing approximately 50 million people live in modern slavery, of which 56% are in forced labour conditions. To put this in context: the individuals affected by modern slavery exceed the populations of Canada, Switzerland and New Zealand combined!
Research by the Finance Against Slavery & Trafficking (FAST) initiative within the United Nations University Centre for Policy Research further said: “to bring this figure close to zero by 2030 – to meet the UN Sustainable Development Goals Target 8.7 – we would need to reduce the number of people affected by around 10,000 individuals per day”. This catalyzed the creation of a public interest consortia, iEARTHS. A group, of which we are a co-founding member, committed to pooling their collective expertise to develop innovative solutions which support eradication of several human rights issues.
Our vision became act now to support a wide range of actors in identifying and eradicating forced labour practices. The vision is simple, impact requires a clarity of focus. The initial focus area chosen, following outreach, is forced labour (including child labour) within international supply chains.
Deciding on the Technology Foundation
Next, the technology approach.
Causal artificial intelligence (Causal AI) was selected as it explains and illustrates complex causes and effects behind any given outcome. It develops its learning from domain knowledge, specific knowledge, and extracts insights from historical data that purely predictive AI often fails to recognize. Achieving the desired vision requires a collaborative approach, one that includes cross functional, cross discipline and cross organization input. The consortia adopted a structured methodology and iterative development cycle with significant human input throughout to validate the base assumptions, inputs, and outcomes during each cycle. Effectively each AI training step includes data input, learning, testing, iteration, and outcome validation (see figure 1).
Figure 1: Collaborative PoC Approach, reprinted with permission from Parabole.ai, 2024
The Inner Workings of Causal AI
The teaching methodology begins with developing hypotheses statements. These statements align with the datasets or data elements needed, effectively establishing specific questions to answer. Hypotheses in our AI analysis are effectively an exercise in identifying an ecosystem of all possible reasons (i.e.: causes) relating to an issue of interest (i.e.: an effect) within a defined setting (i.e.: context). A co-developed framework emerged from this work on extending behaviour analysis (figure 2) to illustrate the interconnectedness of concepts within forced labour.
Figure 2: A Hypothesis Illustrative Example, reprinted with permission from BEworks Inc., 2024.
To contextualize the hypotheses challenge – imagine that you are trying to identify the possible causes (motivators) for a person engaging in a forced labour act when you have no contextual reference or expertise to recognize the environmental conditions within which it may occur. The conundrum of “you don’t know what you don’t know”, a modern version of “I only know that I know nothing” (Socrates).
Addressing these knowledge gaps requires subject matter expertise. This is to ensure the AI learns correctly, in effect preventing the technology from asserting incorrect yet plausible conclusions (aka hallucinations). Expertise is found in different forms including, for example, survivor voices, labour or human rights organizations, or organizations specializing in ethical trade and human rights. Including this very human perspective within AI learning is critical to eliminate biases or the application of values and norms that lack universality.
One such example emerged when we learned that a trafficker can be perceived, relatively speaking, by their victim as “kind” because he/she/they do not abuse them in the same manner as others have in the past – the trafficker’s abuse is often psychological and subtle. This example illustrates some of the human complexities that need inclusion, and appropriate handling, within artificial intelligence training. It was essential that qualified subject matter experts ensured, during AI training, that the right language was used, assumptions challenged as were biases, and proposed outcomes throughout the technology build.
In addition to the human considerations, data privacy concerns and the ethical considerations of data and technology usage are critical and multifaceted. On the data privacy element, for instance, anonymized data is foundational – nothing was used that could lead to identification of a specific person to protect both their right to privacy and security. On the technology side, the consortia considered safeguarding the technology from malfeasance. For instance, using the technology for social profiling or by bad actors to identify new concealment practices.
The vision for iEARTHS is to both support the public interest by empowering NGOs with evidence-based information to support and inform their work, and to support industry in managing legal, reputational, and business risks and opportunities. The “game” changes when data is universally available to industry rightsholders, such as employees, investors, suppliers, customers, governments, and regulators.
Technology is Exciting…Reality Sets In
Beyond the technology, we considered how the solution could become decision useful within industry. For instance, how might it improve business predicative analysis, legal risk mitigation strategies or sustainable decision-making? Exactly what types of business challenges might the technology solution help solve?
Typical business challenges cited included the need to improve an organization’s supplier due diligence verifications (i.e.: lowering legal and reputational risks), establish fit-for-purpose actions to improve business critical supplier labour practices (i.e.: lowering legal, regulatory and reputational risks while increasing long-term cost benefits) or optimizing procurement decisions balancing legislation, environmental and social considerations, company values, cost, time and quality (i.e.: optimizing cost-benefit trade-offs). The vision is that every decision maker becomes empowered to make decisions for which they are more fully informed, and by extension more accountable.
Corporate leaders overwhelmingly want to do the right thing, their question is always how to best do so.
Beyond the obvious market decisions, companies should consider proactively collaborating with technology innovators to pilot AI solutions and help address real world risks and issues (thereby enjoying a first-mover advantage, and unique insight into their own organization). It becomes a symbiotic relationship for both parties when done well.
The Bottom Line
Stay tuned, this forced labour in international supply chains consortia iEARTHS is actively underway. You can get updates through Parabole (parabole.ai), BEworks (beworks.com) or right here!
Let us never forget that the choice to act, or not, lies within each of us…