YACOOB ABBA OMAR: AI’s potential and pitfalls: time to bring in the regulators

The Artificial Intelligence (AI) Safety Summit, hosted by the British government on November 2 at Bletchley Park, home to codebreakers during World War 2, noted that “risks arising from AI are inherently international in nature, and so are best addressed through international co-operation”. 

This was the latest in a series of debates on the effect of AI and its regulation. It was preceded by the September meeting of the Group of Seven countries, chaired by Japan, which committed to the “Hiroshima Process”, aiming to regulate AI and allied technologies. 

This in turn was preceded by the May US Senate hearings where Sam Altman, CEO and founder of Open AI, testified that “if this technology goes wrong, it can go quite wrong”, arguing that “regulatory intervention by governments will be critical to mitigating the risks of increasingly powerful models”. 

Such is the alarm about this rapidly expanding technology that many academics supported MIT professor Max Tegmark when he called for a six-month pause in giant AI experiments.

The Bletchley Park gathering, which included China, resolved to “sustain an inclusive global dialogue and contribute in an open manner to broader international discussions and continue research on frontier AI safety”. 

Job losses, access to privacy and spread of misinformation are among the key concerns researchers, entrepreneurs and regulators have been raising. Leading AI scholar Geoffrey Hinton left Google recently fearing AI could create havoc in society through misinformation and job losses. On the other end business leaders such as Narayana Murthy, co-founder of Infosys, believes in the supremacy of the human mind, arguing that such technologies should be seen as coworkers.

However, it is clear that the service-orientated industries will be the worst hit, especially where developed countries countries have offshored some of their back-end operations. Compared to blue-collar workers, office-bound white-collar workers wedded to their laptops are set to be on the redundancy lists. 

Invite rector

It is estimated that generative AI will have a market value of $60bn by 2025, representing 30% of the total AI market. Most Business Day readers will be familiar with the fact that ChatGPT reached 1-million users in just five days, with new uses of generative AI being launched daily.

Concerned about the implications this seismic shift in technology portends for SA, the Mapungubwe Institute (Mistra) has decided for its upcoming annual lecture to invite former University of Johannesburg vice-chancellor and recently appointed rector of the UN University and UN under-secretary-general Tshilidzi Marwala to address the question of the perils and welfare effects of AI. 

SA has already notched up several milestones, such as the 2020 report of the Presidential Commission on the Fourth Industrial Revolution, Mistra’s “Leap 4.0: African Perspectives on the Fourth Industrial Revolution”, also launched in 2020, as well as the World Economic Forum-sponsored Centre for the Fourth Industrial Revolution Network for Global Technology Governance, which has helped link us to a number of other countries.

The Boston Consulting Group’s recent report, “SA and AI” points out that in the domestic context misuse of AI “could place an organisation or business in breach of its obligations under the Protection of Personal Information Act”, calling for the issues raised and the possible consequences to be tackled by the appropriate policymakers. 

The report also suggests that public-private partnerships and development finance institutions could “embed one further foundation stone of SA’s AI development — tech infrastructure”, allowing data-sharing platforms to build the large, diverse data sets on which AI models are trained. At the local level, especially for remote and underserved areas, stable and high-bandwidth connectivity could be availed full-time through proper investment.

South Africans have already seen the launch of a number of impressive start-ups using AI such as Aerobotics, used in the agriculture sector; Envisionit Deep AI, used by radiologists; and the better-known online marketplace Jumo. 

Marwala’s lecture should provide insights on how the regulatory mechanisms could work with the animal spirits of entrepreneurship in the best interests of SA. 

• Abba Omar is director of operations at the Mapungubwe Institute.