One of the most revealing aspects of the Draft South Africa National Artificial Intelligence Policy published in the Government Gazette on 10 April is the way in which it positions South Africa within global and African frameworks. That is both its strength and its weakness, as it envisages an environment in which the country is less dependent on technology developed elsewhere, while not addressing the absence of a strong underlying foundation for digital inclusion.
Critically, however, it aligns SA policy with global best practice, including the emerging global standard for AI management, ISO 42001, along with the globally recognised NIST AI Risk Management Framework. This alignment, it says, must “emphasise a fair balance between policy-level oversight and organisational-level compliance to encourage safe and transparent AI adoption”.
Coincidentally, a few days after the release of the document, global workforce software company Workday hosted a media conference in Dublin, Ireland, addressing precisely these challenges facing lawmakers in regulating AI around the world. Workday’s own AI management system is certified to ISO 42001 and independently verified to be aligned to the NIST framework. In a panel discussion titled “Globally compliant, locally relevant”, Workday chief corporate affairs officer Chandler Morse in effect endorsed the South African approach.
“We have long thought that smart, workable governmental safeguards around AI would go a long way to building trust and supporting innovation,” he said.
“We don’t see those as mutually exclusive. There’s been this tension between regulation and innovation (and) the story now is a little bit more focused on innovation through regulation. And particularly when it comes to technologies for which trust needs to be a foundational element.”
A message that came though strongly during the conference was that companies like Workday “don’t need government officials to tell us how to develop AI responsibly”. But that didn’t mean governments must stay away.
Said Morse: “We have our own values to drive that. We have our own customer relationships to drive that. But we do see the benefit of a regulatory envelope to put AI in that gives those who will be on the on the receiving end of AI tools some sense that their data is protected, their rights are protected, and they have some sort of transparency and explainability.”
As a result, Workday has engaged extensively with European Union policymakers in the development of the much-delayed EU Artificial Intelligence Act.
“We have views on what we think an AI regulatory framework looks like,” said Morse. “Sometimes that aligns with what governments want to do, and frankly, sometimes it doesn’t. So we end up playing this push and pull of proactively leaning in where we think things are headed in the right direction, and working to moderate some of the proposals that are coming out.”
Asked by Gadget how global AI platforms like Workday manage tension between global standardisation and evolving local regulatory demands outside the EU, Morse said: “There’s a lot of there’s more convergence than there is divergence. The EU AI Act was an early mover and set the framework. On the other hand, the US launched the NIST, an Irish management framework, and we view those as two ends of the pole.
NIST, he said, was voluntary and flexible, while the EU AI Act was a more regulatory focused regime.
In countries like South Africa, Brazil, South Korea, Japan, Canada, Australia, Vietnam and India, policy largely falls in between those.
“There was a knee-jerk reaction of everyone needs their own EU AI act. And that has moderated over the last year. Not unlike Europe, many of these same countries have data protection regulations or laws. The OECD is doing a good job of saying: ‘We’re not going to tell countries what to do, but we really are. If we’re going to see this innovation take hold, we really stay between these yellow lines.’ If you’d asked me this a year ago, I’d have told you I wasn’t sleeping well at night. Now we’re easing into a path where there’s some level of harmonisation.”
Pierre Gousset, vice-president for solutions at Workday, told Gadget: “It’s better to design for a higher bar. So, if you meet the European standards, that means that you’ll be able to meet more or less all of the available standards in the world.”
* Arthur Goldstuck is CEO of World Wide Worx, editor-in-chief of Gadget.co.za, and author of “The Hitchhiker’s Guide to AI – The African Edge”.