At a time when technology is advancing at a rapid pace, the development and deployment of artificially intelligent (AI) systems should not be left to the market’s geopolitical powers alone. It’s time for South Africa to take up its place and offer a unique perspective on the technology, writes Mergan Velayudan, Executive Head: Emerging Technologies at MultiChoice South Africa.
In a world where cross-border trade, knowledge sharing, and information exchange are encouraged, it's crucial to recognise that the decisions made by people on one side of the globe can directly affect others.
So, when it comes to shaping the way AI systems work, especially in solutions intended for global use, the need for diverse perspectives is quickly becoming a necessity.
AI systems are not immune to the biases that exist in human society. Biases inherent in data can lead to skewed outcomes, along with unjust, unequal, and potentially harmful AI decisions. Thus, a more inclusive approach is crucial to creating responsible and balanced AI technologies.
Decolonising AI ethics
Discussions surrounding AI ethics have gained momentum, but the lack of global representation in these conversations is a critical issue that needs addressing.
Historically, AI research has been concentrated in a handful of regions, leaving much of the world underrepresented. This regional disparity is evident not only in the development of AI guidelines but also in the skewed training data used for AI systems.
The dominant positions of the US, Europe, and China in the AI landscape underscore the need to challenge existing power dynamics and foster more equitable distribution of AI benefits. This is particularly crucial for a country like South Africa, where the opportunity to shape AI systems that align with local values and challenges remains largely untapped.
One way to do this is for South Africa not simply to accept this status quo, but to actively engage in major forums on the world stage. Being present in such discussions gives a voice and representation to the country.
To further decolonise AI ethics, it’s necessary to take greater proactive control of the process. Rather than allowing external influences or geopolitical powers to shape discussions and development, this is where investing in education becomes crucial.
We need to build our own research capacity to actively contribute to AI ethics discussions. This requires nurturing a community of researchers, philosophers, scientists, and engineers who can address our country’s unique challenges and perspectives.
For South Africa to be more than a consumer of AI solutions, we need to transition into being a producer. By developing a robust AI industry, we can shape AI applications based on our own requirements and contexts.
Establishing structures at both governmental and private sector levels to encourage AI innovation and expertise within the continent is essential. By actively participating in building AI solutions, we can contribute our unique perspectives, highlight potential biases, and propose ways to rectify them.
Beyond this, private sector initiatives, such as collaborations between universities and companies, play a pivotal role in cultivating local AI talent. These partnerships help create a pipeline of skilled AI practitioners who can drive advancements within the country.
An inclusive approach
As we develop our own AI expertise, we gain a stronger foothold in shaping AI ethics discussions. The values embedded in AI systems are determined by those who build them, making the collaborative effort of establishing ethical principles crucial.
Research has shown that diverse teams make better decisions. Combining viewpoints from different backgrounds, experiences, and geographical locations can lead to more comprehensive consideration of potential pitfalls in AI systems.
Fostering collaborative efforts among diverse stakeholders is also essential for establishing a more universal set of ethical principles in AI. By encouraging dialogue and knowledge-sharing across various perspectives, a broader and more balanced ethical foundation can be laid.
In the journey towards a more inclusive and balanced AI ethics landscape, technology-led organisations, like MultiChoice South Africa, are playing a pivotal role.
By nurturing a homegrown community of experts, we can uncover insights that are more locally relevant. This approach not only improves the identification of biases but also increases the likelihood of mitigating them prior to deployment.
Through our commitment to hiring diverse talents and incorporating viewpoints from across the country and beyond, we’re actively fostering inclusivity. What’s more, the implementation of our AI ethics and governance policies sets a standard for responsible AI development.
By involving engineers, ethicists, legal experts, and business professionals, our AI systems can be ethically grounded and responsible from the outset.
So while the challenge lies in amplifying underrepresented voices and cultivating AI expertise within South Africa, we believe that through proactive involvement and collaboration, a more inclusive and ethically responsible AI future is possible.