I used an LLM to help draft this post and it likely contains >10% AI-generated text, but I’ve edited/rewritten it extensively and endorse it.
On 28 June 2026, we hosted our first Pie & AI Lusaka, Understanding AI Safety in partnership with the DeepLearning.AI community. Pie & AI is a global series of community-led meetups that bring people together to learn about artificial intelligence and discuss its opportunities and challenges.
The event was organised by my co-founder, Lloyd Situmbeko, who serves as DeepLearning.AI Ambassador for Zambia and Cursor Ambassador for Zambia, together with me as Co-founder of AI Safety Zambia. Learn more about our work here: https://aisafetyzambia.com
We welcomed students, educators, creatives, software developers, and professionals who wanted to better understand modern AI and why AI safety matters.
Our aim was to make AI safety accessible. We wanted participants to see that AI safety is not just for machine learning researchers. As AI becomes part of healthcare, finance, education, agriculture, media, government, and everyday digital tools, questions about safety, reliability, governance, and responsible deployment become relevant to all of us.
Building a Foundation
Lloyd opened the session by introducing participants to the fundamentals of modern AI.
Rather than assuming prior technical knowledge, he explained how deep learning works, how neural networks learn from data, and why recent advances in AI have been driven by three key ingredients: data, compute, and algorithms. He also explained how models are trained and why increasing scale has led to the capabilities we see in today's AI systems.
One topic that generated a great deal of discussion was hallucinations. Lloyd explained that large language models predict likely sequences of words rather than verifying facts, which helps explain why they can produce responses that are both convincing and incorrect. This provided an important foundation for the later conversations on AI safety and responsible AI use.
AI Safety Is for Everyone
I led the second part of the session, focusing on AI safety and governance.
One message I wanted participants to leave with was that AI safety extends far beyond technical research. It also includes governance, public policy, ethics, evaluation, standards, and public participation.
We explored ideas such as alignment, robustness, interpretability, and responsible deployment while discussing questions that governments, researchers, and communities around the world are increasingly facing.
A major theme was the importance of African participation. Much of today's frontier AI is developed outside Africa, yet these systems will increasingly shape our economies, education, healthcare, and public institutions. Building local expertise and contributing African perspectives to global AI governance conversations will be essential if we want these technologies to reflect our realities and priorities.
I also shared some reflections from my experience participating in the OpenAI Red Teaming Network and why evaluation, adversarial testing, and governance become increasingly important as AI systems become more capable.
Learning Across Disciplines
One of my favourite parts of the meetup was the discussion that followed.
A graphic designer spoke about AI as a productivity tool while expressing concern that beginners may rely on it before mastering the fundamentals of their craft.
An educator raised questions about students becoming overly dependent on AI and the importance of preserving critical thinking.
A creative writer reminded us that AI safety should also consider culture, governance, fairness, and representation alongside technical performance.
What struck me most was how naturally people from different backgrounds contributed to the conversation. AI safety is strengthened when engineers, educators, creatives, policymakers, and community members all have a seat at the table.
Looking Ahead
The discussion ended with ideas for future workshops on prompting, verification, and responsible AI use. Participants also expressed interest in continuing conversations around AI governance and exploring ways to build a stronger local AI safety community. New attendees were onboarded to our whatsapp community for further learning and resource sharing.
For me, this meetup reaffirmed why we started AI Safety Zambia.
Artificial intelligence is advancing rapidly, yet opportunities to learn about AI safety remain limited across much of Africa. Community events like Pie & AI help bridge that gap by creating spaces where people can learn together, ask thoughtful questions, and discover how their own skills and experiences can contribute to building a safer future.
Not everyone who attends will become an AI safety researcher, and that is perfectly fine. Some may become policymakers, educators, entrepreneurs, communicators, or informed citizens who help shape how AI is used in their communities.
Every conversation grows the ecosystem. Every event builds local capacity. This is the first of many opportunities to help more Zambians and more Africans engage with AI safety and contribute to shaping the future of artificial intelligence.
