25 Nov 2025

DSFSI 2025 Retrospective (Day 2): Beyond the Symbols—A Year of Visionary Leadership

Day 2 of our retrospective: Celebrating Prof. Vukosi Marivate's inaugural lecture and the visionary leadership that shaped African AI discourse in 2025.

Day 2 of our 7-day retrospective celebrating DSFSI’s remarkable 2025 journey.

Beyond the Symbols: A Year of Visionary Leadership

Every research group needs more than technical excellence—it needs vision, voice, and the courage to ask hard questions. In 2025, DSFSI’s leadership didn’t just advance African NLP research; it reshaped global conversations about what AI should be, who it should serve, and how we build it.

🎓 The Inaugural Lecture: Beyond the Symbols

On August 26, 2025, Prof. Vukosi Marivate, Chair of Data Science at the University of Pretoria and leader of DSFSI, delivered his Inaugural Lecture titled “Beyond the Symbols: Natural Language Processing as an Adaptive Problem.”

The Core Argument: NLP as an Adaptive Problem

Prof. Marivate argued that Natural Language Processing must be treated as an adaptive problem—one that requires not only technical innovation but also equity, participation, and cultural relevance.

Global AI systems often overlook African contexts. They’re built on English-centric data, Western cultural norms, and assumptions that don’t translate. The result? Technology that doesn’t understand how we speak, what we value, or who we are.

The solution isn’t to wait for others to fix this. It’s to build homegrown resources, models, and research communities that center African languages, contexts, and people.

A Decade of Impact: 2015–2025

Prof. Marivate reflected on his research journey since 2015:

  • Addressing data scarcity through innovative techniques and augmentation methods
  • Building open datasets like Next Voices—a large-scale multilingual speech dataset covering 7 South African languages with 3,000+ hours of audio
  • Developing adaptable language models like PuoBERTa and Pula
  • Championing open licensing frameworks like NOODL and Esethu to protect community data and ensure equitable benefit-sharing
  • Growing communities and ecosystems through Masakhane, and the Deep Learning Indaba

“NLP must be treated as an adaptive problem — one that requires not only technical innovation, but also equity, participation, and cultural relevance.” — Prof. Vukosi Marivate

Introducing AfriDSAI

The lecture also formally introduced the African Institute for Data Science and AI (AfriDSAI), a new platform to enable breakthroughs in AI and data science, support talent development, and strengthen continental collaboration.

From DSFSI’s strong foundation in African language processing, AfriDSAI expands into health, agriculture, climate, governance, and other critical domains—all through transdisciplinary, African-grounded research.

📺 Watch the full lecture: YouTube Recording (Enhanced with slide overlays, courtesy of UP Faculty of Engineering, Built Environment and IT)

🤖 AfriCHI 2025: Do Africans Want Chatbots?

In November, Prof. Marivate delivered a provocative keynote at AfriCHI 2025, asking a question that cuts to the heart of African AI:

Do Africans really want chatbots?

The talk explored how Africa’s technology future depends on bridging Artificial Intelligence (AI) and Human–Computer Interaction (HCI), bringing together the people who build intelligent systems and the designers who make them human.

Why This Question Matters

While large language models and generative AI dominate global headlines, most systems are built far from the contexts where they’re used. African users interact daily with technologies that don’t understand their languages, cultures, or values.

Prof. Marivate reminded us that AI cannot succeed in Africa if it remains detached from human realities. Machine learning often assumes ideal users and perfect data, but African contexts are multilingual, complex, and grounded in lived experience.

HCI brings the tools to bridge that gap: empathy, co-design, participation, and social grounding.

Activism Through Design

The keynote called for “activism through design”, a mindset where building technology becomes an act of care. AI in Africa must be:

  • Participatory (built with communities, not for them)
  • Trustworthy (transparent, explainable, accountable)
  • Rooted in Ubuntu (our technologies, like our societies, are defined by relationships)

“AI won’t thrive in Africa without HCI , without systems that are trustworthy, usable, and socially grounded. The future belongs to technologies that don’t just speak for Africa, but with Africa.”

Read the full AfriCHI reflection

🎙️ Media and Public Engagement

Prof. Marivate’s leadership extended beyond academic venues:

Nature Careers Podcast

Featured in the Nature Careers Podcast, discussing:

  • Career pathways in African AI research
  • Building research communities in resource-constrained settings
  • Balancing global impact with local relevance
  • Mentorship and the next generation of African AI researchers

This conversation reached global audiences, positioning African AI research as a model for inclusive, impactful science.

Multiple Conference Keynotes and Panels

Throughout 2025, Prof. Marivate shaped AI discourse through:

  • Global AI Summit on Africa (April): Panel on “GenAI Gaps and Opportunities for Inclusive Development”
  • Deep Learning Indaba (August): AI Policy in Africa keynote panel
  • Voices of Africa 2 Workshop (August): Opening address on building NLP infrastructure
  • AfriCHI (November): Keynote on AI and HCI

🌍 What This Leadership Means

Visionary leadership isn’t about having all the answers, it’s about asking the right questions:

  • How do we build AI that adapts to African realities?
  • Do our technologies reflect our values?
  • Are we centering communities in design and development?
  • How do we move from extraction to equity?

In 2025, Prof. Marivate didn’t just ask these questions, he helped build institutions, frameworks, and communities to answer them.

💡 Key Themes from 2025 Leadership

  1. NLP as adaptive problem-solving, not just technical optimization
  2. Community participation as foundation, not afterthought
  3. Ethical data governance as prerequisite for sustainable AI
  4. Bridging disciplines (AI + HCI, NLP + Legal Studies, Tech + Ubuntu)
  5. Building institutions (AfriDSAI) to sustain and scale impact

Tomorrow (Day 3): We explore institutional growth—from DSFSI to AfriDSAI and the partnerships that made it possible.

Related Posts:

Stay connected:

This is Day 2 of our 7-day retrospective. Join us tomorrow as we explore DSFSI’s institutional evolution into AfriDSAI.