Day three of the conference started with a session on “Emerging technologies and the changing world of work.” This session started with an expert talk by Willem Fourie from the Policy Innovation Lab at the University of Stellenbosch. He looked at the promise and peril of using AI to analyse public data. This session covered the genesis of AI systems, he included examples of using AI to analyse qualitative data and the benefits and risks of using AI to analyse qualitative data.
He said: “Policy making is not only about writing policy, it is also about implementing and sometimes ending [policy]. Our institutions need to be ready for AI, we need to support human capability training.”
He took conference guests through an explanation of AI (technology that enables computers and machines to simulate human intelligence and problem-solving capabilities) and how it was inspired by the human body. He noted: “ChatGPT gives wonderful answers but it does not understand what it is telling you. It mimics human intelligence but it has no understanding.”
He covered the ways that AI learns and key ethical considerations and he elaborated on the need to keep data clean so you have reliable data, “It’s boring but very important.”
A panel discussion followed on using AI to analyse data and transform the products and services. Topics covered by panellists included cases of the use of AI in the private sector, the benefits of using AI to transform products and services and risks associated with using AI to transform products and services. Fourie highlighted “the emphasis is on keeping the human in the loop.”
On the panel was Lerato Mathabatha, Public Sector Director, Microsoft South Africa. She explained: “Generative AI has been around for decades, and now we’re at a point where we are putting it in people’s hands. There is a love-hate relationship with AI at the moment but its dependent on the user. It’s here and how do we help users to interact with AI that minimises the risk but leads with its benefits?”
She highlighted the need for prompt training, “If I know how to prompt [AI] and feed it the right information to get the quality of data that is substantial – all of that is on the human.’ She noted the challenge with getting reliable and clean data from the public sector – “our [government] data is paper based and the transition to digitise is slow.”
Thabo Makenete, Head of Public Policy, Southern Africa Region, Meta explained the ways that Meta uses AI in its products and to support government’s work. He explained the AI project Meta implemented with the IEC on the South African national elections to prevent the spread of misinformation.
Stephen Smith, Head, BCB Risk Decision Science, Standard Bank highlighted the emphasis on governance and risk frameworks in the commercial banking space. He said: “All our AI models must consider fairness, bias, ethics and must be implemented on a sound platform. Even if we don’t put in gender, age, race or income the historical data has bias. So we stick to algorithms like logistic regression as we can look inside the model.”
This next session delved deeper into emerging technology and looked at the topic of how digital technologies can enable equitable access to public services with examples from basic education, health and infrastructure. The panellists shared some fascinating case studies of how technology is enabling access to educational resources and healthcare.
Mark Horner the founder of Siyavula, a free platform focused on mathematics and physical sciences shared this impressive EdTech solution and shared how it is impacting learners in South Africa. He highlighted that their online textbooks are read by 50% of maths and science learners in South Africa and that this is made possible through zero rating of content by cellphone providers. “We can get to the majority of high school learners today if we leverage mobile phones.”
He finished by saying: “There are 100 EdTech companies in South Africa but there is no policy, we all have data but no way to share it with the Department of Basic Education. If you do have data we can make systemic changes.”
Next William Mapham, Founder of Vula Mobile presented his software that connects health workers to improve education outcomes. He shared that the Vula app has 37 000 healthcare workers and that they have helped 1,9 million patients. The app uses digital records, chat data and images to create a full patient profile. “We improve health outcomes and there is time and money saved for healthcare workers.”
The final presentation was from Phemelo Mahamuza: Gauteng City Region Observatory, she shared the infrastructure reporting model to track health facility construction. The model is used to determine if the infrastructure there, is it ready to use and how long did it take to build? This was initiated by the Department of Health to validate the construction of health care facilities using remote sensing and machine learning to classify land cover.
The next session looked at administration data for research and policymaking and featured researchers working at National Treasury’s Data Lab using tax data to inform policy and planning.
Bruce McDougall from UNU-WIDER shared information on the National Treasury’s Data Lab showcasing the data available and how the National Treasury is working with researchers to understand the implications of policy changes and to inform policy changes. He highlighted the benefits of using administrative data and also the challenges of this data including that “the downside is it’s not designed for research so its messy, little documentation, little control what is captured.”
Marlies Piek, Senior Economist, National Treasury said: “Many important policies are implemented based on household surveys and we don’t have accurate information, an example of this is the national minimum wage.”
She shared some case studies where this data has been used to reflect on the impact of policy reform including the mixed results of the introduction of the youth wage subsidy, that the introduction of the marginal tax rate actually resulted in a reduction in tax revenue due to a behavioural response from high-income earners. Finally she shared the impact of the agricultural minimum wage and that employers didn’t let go of staff but they stopped hiring new workers.
She concluded by saying: “I think we have showcased that data can be shared across institutions in a safe way and there is power in sharing data, it can improve the lives of citizens.”
Finally Andrew Nell from SEAD SA showcased the digital tool that has been created using this data to provide data at a municipal level. He showcased the tool and highlighted that municipalities are using the tool to review economic activity and performance and the impact of disasters like Covid-19, flooding and riots.
This was followed by a lively debate between the students attending the conference. Ronette Engela: Acting Head, GTAC commented: “The energy and insight that you bring as youth is fantastic. Professional colleagues, lets take our leadership from the front benches where the energy is up. I hope you have found value in this conference.”
The conference conclude with a video message from Ashor Sarupen: Deputy Minister of Finance. He said: “This topic is timely and crucial for the growth and development of our country. The sessions have demonstrated the breadth and depth of opportunities available in digital technologies. We see how innovation can drive efficiency and effectiveness in government service delivery, as well as the collective effort required to drive transformation. We must work together. The knowledge shared here will shape policies and strategies. I urge you all to take the insights gained here back to you institutions. Together we have the power transform the lives of millions of South Africa.”
If you missed the sessions you can download all the presentations here or catch the recordings on YouTube.