There is no more denying the fact that Artificial Intelligence is reshaping the world in more ways than we could have possibly imagined. For example, AI could detect cancer much earlier and with increased accuracy. Moreover, it is projected that by deploying AI, businesses will save up to $8Billion annually by 2026. Nonetheless, in the midst of these impressive achievements are concerns that Africa must take note of. As a continent that is not playing a significant role in shaping the conversation on AI, albeit the conversation being ongoing, Africa must work with partners to ensure that inclusivity and representation are prioritized as early as in the Development of AI models. Unless this happens, we might wake up one day to realize we did not do enough and by then, it will be too late-as put in 2023 by OpenAI CEO, Sam Altman “If the technology goes wrong, it can go quite wrong.”
While this conversation takes shape, there are two prominent perspectives that standout in relation to Africa; 1) how the continent might or must reposition itself to benefit from the AI revolution on one hand, and 2) the need to build effective AI governance on the other. The World Bank drills deeper, talking about fostering equitable access to AI, and aligning the discussion around AI with sustainable Development Goals. Whereas both perspectives are essential, this essay contends that Africa must also work to ensure that inclusion and representation particularly in the development stages of AI are prioritized.
As we speak, the United States and China standout as de facto leaders in the global AI revolution. However, some commentators speculate that the competition between the two could follow the same trend as the cold war era arms race between the United states and the Soviet Union. To recap, this race resulted into unprecedented proliferation of nuclear weapons with the two great powers creating ever more destructive capabilities. On the contrary, many countries did not develop nuclear capabilities at the same pace or at all which made the nuclear threat even more profound. In the same way, the current trend in AI risks creating a similar scenario, especially in the absence of effective AI governance frameworks. For example, Trump era Secretary of Defense, Mark Esper once pointed out that, “…which ever nation harnesses AI will have a decisive advantage on the battlefield for many years.” This goes to show how nations might deploy AI in the absence of such frameworks, ways that might jaundice efforts to build a peaceful, fairer and equitable world.
Experts who have been bold to express their concerns about AI have highlighted the emphasis placed on “AI for good” Vis-a-Vis Ethics of AI in the global conversation on Artificial Intelligence. In fact, it is easy to come to the consensus that, AI might not have a problem, but rather how we use it and where we use it. Moreover, one area often cited as a source of potential risk is, that of training data. The key concern here has been, most of these models have been implemented prior to the establishment of any frameworks for AI governance. Thus the question becomes what is the quality of training data used? And how well is Africa represented in these datasets? With almost all of the training datasets coming from the global north, there is a significant risk of reinforcing existing stereotypes, and power dynamics which have for a long time held Africa at a disadvantage. Moreover, in an age where nearly 60% of online content is produced by AI, subsequent improvement of AI capabilities could only exacerbate this dynamic since it is based on already biased input.
The risk of accelerated bias: Different Tech enthusiasts and experts have on numerous occasions highlighted the possibility of AI creating abundance on the bright side. Nonetheless, there have been warnings from others, regarding the risk of accelerated bias, particularly in cases where the initial training data was prejudiced or not representative. Thus, the Cinderella issue in the AI conversation is that of representation, and quality of training data. Indeed, algorithmic biases directly affect both fairness and trust, jeopardizing any benefits presented by AI. Given the historical injustices and cultural sensitivities on one hand, and the bulk of training data coming from the global north on the other, AI models could aid in perpetuating historical injustices, stereotypes, thus undermining uptake and subsequent access to the benefits of AI technology on the continent. This is one particular reason why Africa needs to rise to the occasion and play its due role in this conversation.
As China and Africa made renewed commitments to jointly building an all-weather China Africa Community with a shared future for the new era, both partners agreed to work together in order to seize the historic opportunity of the new round of technological revolution. This commitment made at the 2024 FoCAC summit in Beijing aligns well with the need to address the inclusion and representation in the development of AI. The effort will seek to foster an inclusive, fair, just and nondiscriminatory environment for the development of science and technology. More so, in the first steps, Africa overwhelmingly welcomed China’s Global Initiatives on; AI governance and Data security. The initiatives are a step in the right direction in as far as promoting the rights of Africa in the global governance of AI are concerned. Therefore, working with China, Africa can enhance not just its role but also representation in novel AI models; from Large Language Models to Generative AI Models and everything in between.
In conclusion, as the universal application of Artificial Intelligence is cascading around the world, Africa can position itself to reap the enormous benefits presented by this revolution. Faced with an absence of AI governance frameworks in most countries, and the risk of accelerated biases, Africa must harness positive partnerships in order to counteract the perpetuation of old stereotypes, injustices, underrepresentation while enhancing fairness and inclusion as inroads to the positive benefits of Artificial Intelligence. In addition, the partnership with China could be a strong advocacy voice for responsible and ethical AI in order to minimize associated security threats and the concentration of power thus buttressing global order and peace in the age of AI.
The author is a research fellow at the Development Watch Centre.
