Upskilling Government Officials and School Leaders in AI

Navigating the educational AI landscape with literacy, fundamentals, and thoughtful policy at the 2025 DBE Lekgotla

I’ve spent quite a bit of time thinking about how AI is fundamentally changing our approach to education. It’s not a distant future technology; it’s here and expanding quickly. As government officials and school leaders grapple with these rapid changes, I’m convinced that proper upskilling in AI literacy and fundamentals isn’t just beneficial — it’s essential.

The author at the DBE Lekgotla 2025

Why Upskilling in AI Matters Now

The pace of AI adoption in education has been rather extraordinary. Most schools are already reporting some form of AI use in the past year, either formally or informally. This widespread adoption has profound implications for teaching, learning, assessment, data privacy, and equity.

Without fundamental AI knowledge, leaders risk underestimating how these technologies can transform teaching and learning. Perhaps more concerning, they might overlook pitfalls that could deepen inequality rather than address it. The cost of not upskilling — measured in poor decisions and missed opportunities — is potentially quite high.

Upskilling helps us be proactive rather than reactive, ensuring fair and effective use of AI in our educational systems. When leaders understand these tools properly, they can make decisions that respect ethical and educational objectives, balancing the need to foster innovation with safeguards that protect students, teachers, and institutional integrity.

Building AI Literacy: The Foundation

AI literacy means knowing how AI models are trained and why they behave in specific ways. I find it helpful to think of AI as a student learning from examples: the more diverse examples it sees, the better it performs. But unlike a student, AI doesn’t comprehend what it learns; rather, it identifies patterns which it uses to make predictions.

While generative AI (like the large language models powering chatbots) is relatively new and driving much of the current discussion, AI technologies have been around us for some time. In education, there are already adaptive learning platforms, AI tutors, and automated grading systems in use. School leaders who understand these systems can better evaluate new tools and embed principles for safe AI into policy and procurement processes.

Understanding AI Fundamentals

As impressive as AI systems appear, they don’t comprehend things in the same way humans do. Instead, they make predictions based on patterns they’ve observed in their training data. In most cases these predictions are correct, but sometimes they “hallucinate” and present made-up information. Because AI can be confident but wrong, human oversight is essential.

Leaders should prioritise training teachers and staff on fact-checking and verifying AI outputs. Additionally, AI is quickly evolving beyond text; multimodal models now handle images, video, and voice. This shift offers new opportunities for teaching — such as analysing student-drawn diagrams or assessing pronunciation in language learning — but also presents new complexities for policy regarding copyright, privacy, and cultural sensitivity.

Key Policy Considerations

I’ve compared policy approaches across Australia, Japan, New Zealand, South Korea, and the UK. Despite their diverse contexts, five consistent themes emerged:

First, all education ministries explicitly treat uncredited AI content as plagiarism while encouraging transparency in allowed use — students must declare AI assistance. Second, all policies highlight the need for ongoing teacher training to identify AI misuse, design AI-resilient assessments, and leverage AI in teaching practice. Third, privacy is a major concern, with all ministries prohibiting uploading personal student data to public AI services.

Fourth, AI tutors in government schools are increasingly seen as potential equalizers that can provide similar benefits to expensive private tutoring. Finally, there’s growing recognition of the importance of AI that can accommodate local languages and cultural contexts to avoid marginalising students (particularly relevant for multilingual countries).

Practical Next Steps for Student Use

All students will have access to high-reasoning AI models on their phones in the next year. It’s rather pointless to try and ban this technology. Instead, I recommend encouraging “transparency by design” — requiring students to cite or declare AI use, treating AI like any other reference source.

AI detection software is imperfect and generally discouraged — it can flag genuine student work (often by second-language English speakers) or fail to catch AI-generated text. Many school systems are now focusing on assessment redesign, moving away from traditional take-home assignments toward in-class, oral, or project-based tasks where students can demonstrate authentic work.

The research is showing that AI can enhance learning when properly integrated — helping students understand complex concepts, providing research and brainstorming assistance, and giving personalised feedback at scale.

Professional Development for Teachers

When it comes to professional development, all teachers should have basic AI literacy — an understanding of AI fundamentals, ethical considerations, and how to evaluate AI outputs for bias or inaccuracy. Support needs to be provided on how to meaningfully use AI in lessons and update pedagogic approaches as needed.

One practical way to do this is to provide ready-made prompts that teachers can use directly to generate resources and reduce their workload while safeguarding academic integrity. A great way to learn about AI is by using AI. The best approach is to implement pilot programs in schools or districts following a three-step process before attempting a broader national rollout.

Privacy, Equity and Inclusion Considerations

Any new technology presents both challenges and opportunities for privacy, equity and inclusion. Some key strategies include vetting all AI tools for privacy, bias and pedagogical value — ensuring they adhere to local data protection laws. It’s essential to explicitly ban uploading personally identifiable information or sensitive data to public generative AI tools.

AI requires proper infrastructure, and policymakers must ensure internet connectivity, devices, and teacher support for all schools — whether rural or urban. AI is trained on large data sets that can reflect biases, so regular audits should be done to avoid perpetuating harmful stereotypes.

I’d recommend AI solutions that respect cultural nuances and avoid insensitive content. There are options available for safer AI chatbots, or you can customise open systems to local curriculum and values. On the positive side, AI has shown much promise for personalised support for students with varying disabilities or learning challenges — an area worth championing.

Measuring Impact and Next Steps

For any project introducing new technologies into a complex system, measuring impact is crucial. We need to establish clear metrics for success, including increases in student achievement, teacher confidence, and learner engagement. Both qualitative and quantitative data should be collected regularly against these metrics, with insights brought into review cycles to assess implementation progress.

My recommendations for next steps include:

Publishing national or provincial guidelines that clarify permissible uses of AI in classrooms and require transparency for AI-generated content. Moving toward “AI-aware” assessment formats that value critical thinking and authentic engagement over memorisation. Providing ongoing professional development focused on AI literacy, ethical usage, and data protection. Enforcing strict data privacy measures and contractual obligations for vendors. Ensuring policies address resource gaps, especially for rural, underserved, or differently-abled learners. And involving teachers, students, parents, and industry partners in shaping guidelines.

Looking Forward

AI is here to stay, and we must be proactive. By upskilling government officials and school leaders in AI literacy and fundamentals, we can navigate both opportunities and threats more effectively. Working together, we can shape a responsible and inclusive AI future in education, ensuring that all learners benefit from these powerful new tools.

The sooner we begin this journey of understanding and implementation, the better positioned we’ll be to harness AI’s potential while mitigating its risks. Education has always evolved with technology — this is simply the next chapter in that ongoing story.

Here’s my slides from the DBE Lekgotal 2025:

[embed]Upskilling_AI_Cambridge.pdf Edit descriptiondrive.google.com