The challenges and opportunities presented by artificial intelligence (AI) have become a crucial topic of discussion. A particular focus is on algorithmic bias, which poses risks of perpetuating inequalities across various communities worldwide. Algorithmic bias arises when AI systems, typically driven by machine learning, produce biased outcomes because the data they depend on is incomplete, imbalanced, or not fully representative.
In this context, a group of researchers from Cambridge and Warwick Business School have introduced a ‘relational risk perspective’ to address these issues. The new approach considers both the current and potential future uses of AI globally, aiming to protect the benefits of AI while minimizing the risks involved.
The impact of AI in the workplace is already evident, affecting jobs, tasks, and social dynamics among workers. According to the researchers, an over-reliance on AI may undermine professional expertise and critical thinking, resulting in human workers becoming demotivated as they defer to AI-generated decisions. This situation not only influences individual tasks but also affects workplace relationships and organizational interactions.
Moreover, the team highlights concerns about the AI industry's reliance on 'invisible' workers in the Global South. These workers are responsible for data cleaning and algorithm refinement and are often excluded from the benefits of the rapidly growing AI sector primarily enjoyed by users in the Global North. This phenomenon, described as 'data colonialism', echoes and reinforces existing global inequalities.
The research underscores the urgency of addressing these challenges as AI becomes more embedded in society. Noting the swift pace of AI's evolution, the group calls for the alignment of ethical and regulatory frameworks with technological advancements.
The relational risk perspective outlined by the researchers does not label AI as inherently beneficial or harmful. Instead, it is seen as having the potential for both, dependent on development and application across varying social contexts. The risks of AI are seen as evolving with the changing dynamics between technology, users, and society at large.
The team calls on policymakers and technologists to proactively address how AI might further entrench or challenge existing inequalities. Acknowledging varying rates of AI maturity among countries, the researchers advocate for the inclusion of diverse perspectives in forming AI risk policies, urging a multidisciplinary approach to mitigate bias and ensure policies reflect the interests of varied communities.