Artificial intelligence (AI) holds immense potential, but it requires reflection on its uses and diligent risk management. This echoes the enduring tension between innovation and regulation.

The tragic loss of life aboard the Titan submersible at the Titanic wreckage serves as a poignant metaphor for the AI era, where the allure of greatness interweaves with danger. The passengers of the Titan were enticed by the promise of awe-inspiring sights, but the risk-taking of an unregulated venture resulted in their needless demise.

The tragedy is a reminder that we should not ignore expert warnings and downplay safety on the altar of progress and innovation.

AI is anticipated to bring about substantial economic growth, with Goldman Sachs, McKinsey, and PwC projecting an increase in global GDP by as much as 7% or a staggering $7.3-trillion.

Despite these prospects, governments worldwide, both in developed and developing nations, along with AI experts, have raised concerns about the risks and proposed regulatory measures.

In contrast, South Africa’s political leadership has only generically acknowledged the potential benefits of AI while giving little attention to its multifaceted risks. In doing so, they inadvertently prioritise innovation while overlooking the hazards.

Apathy of the political class

In South Africa, the regulatory void is compounded by the lack of evidence that policymakers, legislators, or political parties fully appreciate the epoch-defining nature of AI. Several points highlight this relative disinterest.

Firstly, there is a distinct lack of focus on AI among top officials and within policy documents. The most recent mention of AI in the State of the Nation Address dates back to 2019. While the Minister of Communications and Digital Technology, Khumbudzo Ntshavheni, has sporadically expressed critical reflections on AI, there is no evidence indicating that these points have translated into concrete policy initiatives.

Notably, AI remains untouched within key government policy documents such as the National Development Plan. Secondly, AI-related matters have received limited attention in Parliament. The term “artificial intelligence” has only been mentioned in passing throughout the current year and in 2022, with minimal substantive engagement on the subject.

Lastly, major political parties have mostly failed to develop meaningful AI policies. The African National Congress, in its 120-page apex policy document, makes only fleeting references to AI.

Similarly, the policies of the Democratic Alliance do not address the technology, although some officials occasionally comment on the issue.

More encouragingly, the Economic Freedom Fighters has at least made AI-linked proposals in two of its policy documents from 2019, but not too much beyond that.

South Africa’s indifference towards AI positions it as an outlier among upper-middle-income nations. This is exemplified by the country’s low positioning in the AI Government Readiness Index — a global, multidimensional index that considers factors such as government strategy, vision, and governance.

In the 2022 index, South Africa was placed at a disappointing 68th out of 181 countries, trailing behind our BRICS counterparts and many other emerging markets. Further putting the country apart is an absence of a national AI policy, in contrast to African peers like Rwanda and Mauritius.

Political will needed

My research found that AI companies in South Africa are themselves in favour of regulation as it would establish a level playing field and ensure responsible practices.

However, regulating AI presents at least two interrelated challenges.

Firstly, the “pacing problem”. Digital technologies tend to develop faster than the social structures, causing a gap between technology and regulation. A case in point is ChatGPT’s fast-paced disruption of the education sector.

Secondly, the so-called “Collingridge dilemma”. It is difficult to predict and control the impacts of a technology at its early stages, but by the time its impact becomes clear, it is too deeply embedded in society, making it challenging to control. For example, social media’s entrenchment in society, along with all its harms, within a regulatory vacuum.

In South Africa, it is worth considering a third factor contributing to the lack of AI regulation: the apparent absence of political will. This can be partly understood as policymakers prioritising pressing socioeconomic concerns that directly impact citizens’ lives, as AI policies may not be seen as winning votes.

A similar argument can be made on climate change, which hardly registers as a concern among South Africans. However, despite this, the government has shown political commitment to address climate change, indicating a strategic decision to prioritise an important issue despite limited immediate political pay-off.

Our political establishment should also be concerned about AI. Why? While there are multiple risks associated with this technology, a crucial point is its general-purpose nature. AI has the capacity to enhance human cognitive abilities on a large scale and at a remarkable speed, all at a relatively low cost. This means that it can, for instance, help benevolent actors achieve the UN Sustainable Development Goals.

But it can also be used for a near-endless range of malicious acts, such as creating misinformation, committing fraud, and designing weapons of mass destruction. Not to mention unintended consequences like perpetuating social bias, displacing jobs, and exacerbating inequality.

Regulation for innovation

In a democratic system, the responsibility of regulating AI should rest with elected officials, rather than leaving it to technology companies, the existing default position. Moreover, officials must develop a clear vision for AI that encompasses the protection of rights, establishment of responsibilities, implementation of safeguards, and provision of recourse mechanisms.

Having regulations and a well-defined strategy for AI is not a cure-all solution, but they are crucial steps towards governing AI in a manner that benefits society as a whole. This becomes even more significant as AI is increasingly adopted by state institutions, as seen with the utilisation of AI by the South African Revenue Service.

Government must also consider the potential winners and losers that may arise as a result of the technology’s advancements. Policymakers can work towards ensuring that the benefits are distributed equitably and that measures are in place to support those negatively affected by AI. These efforts will take time and will be ongoing as the technology evolves.

In the near-term, government should leverage the Artificial Intelligence Institute of South Africa to help inform the debate, and agencies such as the Information Regulator should enforce laws on data collection, protection, and use in relation to AI.

South Africa’s political establishment should heed the warning of the Titan tragedy — unbridled innovation can come at a grave cost. Whereas regulation accompanied by a compelling vision can spur economic growth, mitigate risk, and promote societal benefits. DM