Missteps Uncovered in South Africa’s National Artificial Intelligence Policy
Technological advances should prioritize the public interest rather than yielding to corporate profit motives, asserts Tyrone McCrindle. Following the recent retraction of South Africa’s draft National Artificial Intelligence Policy by Communications and Digital Technologies Minister Solly Malatsi, McCrindle emphasizes the need for transparent management of AI systems over unchecked corporate influence.
On Sunday, Minister Malatsi withdrew the controversial draft policy after it was discovered to contain at least six AI-generated citations, commonly referred to as “AI hallucinations.” Marazzi noted, “The most plausible explanation is that the AI-generated citation was included without proper verification. This should not have happened.”
Malatsi deserves recognition for acting swiftly in response to the serious errors within the policy. However, the presence of fictitious references in a national document raises significant concerns about public trust. This incident illustrates the potential hazards of relying excessively on digital technologies, highlighting the need for thorough human oversight and proactive public engagement with these tools.
Over the past two decades, digital technology has increasingly influenced our day-to-day lives. AI algorithms play a pivotal role in determining access to jobs, credit, and essential public services. While these technologies have the capacity to enhance participatory democracy and spur socio-economic growth, they simultaneously represent new fronts for political and military power, often leading to dire global consequences.
The competition among major tech companies is cutthroat, with many investing hundreds of billions of dollars into AI development. As they jockey for dominance, the repercussions for the environment and the economy are becoming evident. Reports indicate rising job losses linked to AI in the United States, with companies like Amazon, Oracle, and Meta laying off thousands. Microsoft has announced plans to release more than 15,000 employees in 2025, acknowledging the abrupt impact on those who have dedicated years to the organization.
Amidst these challenges, Google’s negotiations with the U.S. Department of Defense sparked an open letter from over 600 of its employees, urging the company to harness AI for the greater good rather than employing it in harmful applications such as lethal autonomous weapons or mass surveillance. The intersection of technology and social media continues to shape political discourse without sufficient democratic accountability, while digital barriers widen due to unequal access to connectivity and technological skills.
Earlier last week, I reached out to Minister Malatsi on behalf of Article One, requesting an extension on the comment period for the draft AI policy. My initial correspondence outlined five compelling reasons for an extension, including the complexity of the policy, the necessity of multi-stakeholder engagement, and the obligation for meaningful public participation mandated by the constitution.
Shortly thereafter, I learned that a fictitious academic journal was among the references in the policy. Upon further scrutiny, we identified six fabricated sources, indicating reliance on an extensive language model to draft the document. When my request for additional time to review the policy went unanswered, I stressed to the minister the administration’s responsibility to provide accurate information to the public, especially concerning disruptive technologies.
Moreover, the draft fails to address specific harms and solutions, such as jurisdiction over multinational technology companies. This oversight is particularly pertinent, given the ongoing legal dispute between South Africa’s Information Regulator and tech giants Google and Meta, who contend that local laws do not govern their operations despite their significant influence over the personal data of millions in South Africa.
I have urged the Minister to share the 32 submissions received on the 2024 AI Policy Framework, including input from major players like Microsoft and Huawei, so their influence can be assessed in relation to the policy. Our advocacy must focus on AI policies rooted in constitutional values and grounded in inclusive public dialogue, positioning us to effectively address the challenges posed by global technological and corporate dynamics.
The crucial lesson from this episode extends beyond the inherent risks of AI; it serves as a reminder of our collective duty to ensure that technological advancements serve the public good rather than merely corporate interests.
Tyrone McCrindle is the Executive Director of Article One, a public interest organization dedicated to fostering a culture of participatory governance in South Africa by holding governments, businesses, and powerful entities accountable.
The views expressed are not necessarily those of GroundUp.
