Publications
Partners
South Africa lacks a comprehensive regulatory framework for AI, leaving organisations that develop or deploy AI systems vulnerable to ethical risks and public scrutiny. This raises a critical question: How can organisations ensure ethical, responsible, and trustworthy AI usage in the absence of regulation and laws? Writes Emile Ormond.
Artificial intelligence (AI) is rapidly reshaping the landscape of business and society, presenting a wave of possibilities and challenges. There's been a rapid uptake, including by business, of generative AI tools such as ChatGPT.
In South Africa, where the AI industry is growing quickly and AI adoption is on the rise in sectors like health, education, agriculture, and security, there is a pressing need for effective governance and management of AI ethics. This is in the context of AI presenting great opportunity but also risk, which ranges from the mundane to monumental.
Unlike some countries and regions which are well on their way to enacting comprehensive AI regulations, South African organisations face uncharted territory in the absence of legal frameworks.
The country lacks a comprehensive regulatory framework for AI, leaving organisations that develop or deploy AI systems vulnerable to ethical risks and public scrutiny.
This raises a critical question: How can organisations ensure ethical, responsible, and trustworthy AI usage in the absence of regulation and laws?
A study that consolidated insights from local AI practitioners and related experts' attempts to shed light on this question. This article explores some of the findings, along with strategies and principles that South African organisations can adopt to navigate the AI ethics landscape responsibly, upholding societal trust amidst ongoing technological transformation.
The experts flagged several observations with regards to the structural deficiencies of the regulatory environment in South Africa, and some thoughts on how to correct it.
Self-Regulation Insufficient:
There is scepticism of self-regulation by organisations at either an industry or individual company level. Self-regulation raises questions about inherent biases and potential conflicts of interest. However, regulation does not need to be uniform across sectors, rather it should be based on sector-specific risk factors. For instance, scrutiny is most appropriate in already well-regulated areas, such as financial services and healthcare.
Government Oversight Required:
The prevailing view is that the state, in some capacity, should be responsible for regulating AI. However, doubts exist regarding the South African government's capability and resources to effectively undertake such a role. Nevertheless, formal AI governance must involve a multi-stakeholder dialogue, incorporating perspectives from enterprises, government, civil society, and the public.
Existing Governance Code Insufficient:
The current corporate governance framework, such as the country's preeminent corporate governance code, the King Code, lacks the specificity needed to address AI ethics risks effectively. The King Code should be updated to reflect the specific risk challenges of AI.
International Obligations:
South Africa, despite an absence of a local regulatory framework, is still nominally impacted by global non-binding agreements like the UNESCO Recommendations on the Ethics of AI and the OECD Principles on AI. At an African level, regional bodies have yet to establish AI or data management obligations, but these will be applicable once they are in force.
These findings illustrate the complex and evolving domestic and international environment in which organisations must navigate the AI space. However, the experts also identified several internal measures that organisations could reflect upon and institute to help facilitate responsible AI usage.
Awareness of Ethics:
As a starting point, organisations should be aware of and recognise that AI ethics is a critical issue that requires top management and governing board attention. In other words, awareness is a precondition for an organisation to manage AI ethics effectively.
Bottom-Up Consultation:
Organisations should adopt a bottom-up, consultative approach with stakeholders. In other words, they should engage directly with those who use and are affected by the technology and not simply impose it on them from the top-down.
Diverse and Informed Staff:
Foster diversity in the workforce and leadership, incorporating varied disciplinary perspectives and backgrounds (e.g., age, gender, race, and ethnicity), to make more informed ethical decisions. Moreover, leadership needs to be more astute in understanding the technology and its socio-technical nature and consequences. In other words, it is not merely something that the tech or compliance staff should care about.
Develop Existing Frameworks:
Local organisations have access to a number of AI ethics frameworks. However, all of these frameworks were developed by entities in the Global North, including big multinational consulting firms and large technology companies. These frameworks cannot simply be cut-and-paste into the South African context, but rather needs to be tailored to local requirements and conditions, which accounts for the country's unique risks.
Tailored Approach:
Organisations should recognise that AI governance is not a one-size-fits-all approach and ethical concerns and dilemmas may differ depending on a variety of factors such as industry, size, operating model, maturity level, and organisational culture. Each organisation should adopt a customised approach to governance.
Expand Existing Structures:
Integrate AI ethics governance into existing organisational structures, policies, and dynamics. This means leveraging the organisation's existing mission, values, governance structures, and risk management tools and processes.
· These measures, while not being exhaustive, provide a starting point for organisations to navigate AI adoption responsibly and ethically in South Africa's regulatory vacuum.
In the upcoming months and years, discussions on AI regulation will likely gain momentum, but in the absence of formal requirements, organisations must take proactive steps to navigate the AI landscape responsibly.
By prioritising ethical AI practices, these organisations can establish trust, protect societal interests, and harness the transformative potential of AI responsibly.
An ethically-focused approach is not only morally imperative but also offers a strategic competitive advantage. Embracing responsible AI usage allows organisations to lead the way in the quickly evolving technological landscape, positioning themselves as ethical leaders in their respective fields and the myriad of business benefits that flow from this.
