How Malik Afegbua is using AI to rescue Nigeria’s fading oral histories

In Ikorodu, a suburb of Lagos State, Nigeria, a local chief told Malik Afegbua that cultural stories were not meant to be shared. But Afegbua, who was gathering stories for an Artificial Intelligence archival project, saw the danger in the silence. To Afegbua, telling our own stories is not just a choice; it is the only way to ensure that the global tech narrative doesn’t leave the African truth behind. 

Convincing guardians of culture that preservation requires sharing, not secrecy, captures the essence of Afegbua’s current mission. 

At 38, the University of Surrey, England graduate who once sold T-shirts while on campus, is now racing against time to do something unprecedented: create “digital twins” of Nigeria’s elders so that future generations can interact with their ancestors.

From fashion to the future

Afegbua’s journey to becoming one of Nigeria’s most interesting AI practitioners wasn’t linear. In 2008, he started a clothing line while in university. “Nothing in the media space,” he says, “but somewhat in a creative fashion space.” 

After returning to Nigeria in 2010 and completing the National Youth Service Corps (NYSC), Nigeria’s mandatory one-year service program for graduates, in 2011, he fell in love with storytelling.

By 2012, he had launched Sweet TV, an online television platform, years before it became commonplace. 

“I did a lot of things with people who are celebrities today. They were not celebrities back then,” he recalls. 

The platform featured interviews, freestyle sessions, and events that captured Lagos’s emerging creative scene. 

Afegbua has long moved on from Sweet TV, although it remains an archive of his early career on YouTube. 

But even while producing reality shows for DStv’s Spice TV and shooting commercials, Afegbua was always looking ahead. 

“I’ve always been a futurist,” he says. “Even within the media and broadcast space, I was  always looking to the future to understand how we will be living years to come, in terms of information consumption, communication, marketing, everything.”

The problem nobody is talking about

What Afegbua discovered in his AI experiments troubled him deeply. “When you look at AI systems, there is a lot of misrepresentation and bias because our data was not captured properly,” he explains. “Even the ones that are captured are misrepresented data.”

The problem is systemic. When Nigerians talk about documenting culture, they focus on the three major ethnic groups: Yoruba, Hausa, and Igbo, or whatever pop culture exists online. 

“What about the people who are not spoken about? What about the cultures that are not documented?” Afegbua asks. “We have a lot of languages in Nigeria. We have a lot of cultures, sub-cultures. What happens to them? What happens to their stories?”

His answer is Legacy Link, a project that sits within his company, Sleek City.io (the AI arm of his broader Sleek City Media). 

The initiative documents oral histories from Nigerians aged 80 and above, capturing what life was like in their youth, correlating it with today, and projecting what life might be like 60 to 70 years from now.

“I call them the most valuable people for us because they are on the last lap of their life,” Afegbua says. “When they eventually die, it’s like a library being burnt.”

But Legacy Link goes beyond simple documentation. Using the collected data, Afegbua is training large language models (LLMs) to create digital twins of these elders. 

The vision is audacious: future generations will be able to ask their ancestors questions and receive responses based on their actual knowledge, experiences, and worldview.

“For the first time in our lifetime, our ancestors will be available for the next generation to interact with,” he says. “This has never happened before.”

The technical challenge of  cultural memory

Creating digital ancestors is not simple. One major hurdle is AI hallucination—when models generate false or misleading information. 

For a project rooted in cultural preservation and historical accuracy, this has the potential to have catastrophic outcomes. 

Afegbua’s solution involves training custom datasets with guardrails. By restricting the model to specific, verified datasets, he ensures it only iterates on facts. Even when the IA leans into hallucination, the guardrails force honesty. “When you have guardrails on a model system, there is a way it won’t stray too far off,” he explains. 

“Even if the AI is hallucinating, it’s still beating around the same subject or topic, just on a different variation.” 

The system can distinguish between fact and fiction, clearly labeling when it’s extrapolating versus reporting documented knowledge.

The Ikorodu project, to collect cultural data from its custodians, was done in collaboration with the IGA Nigeria Development Lab, a Lagos State parastatal, and the United Nations Educational, Scientific, and Cultural Organization (UNESCO), exemplifying his approach. Working with translator support, he and his team trained traditional chiefs and traditionalists—many in their 50s, 60s, and 70s—on AI and content creation. 

These are people who don’t speak English, yet Afegbua convinced them that their participation in documenting their heritage with technology was essential to its survival.

He conducted 3D scans of artifacts, tagged and documented them, ensuring that when stories are told, “it’s accurate, it’s valid, so nobody else can tell the story for us or for them.”

Distribution: Museums, holograms, and public libraries of the future

Afegbua envisions multiple access points for this cultural database. There will be museums, there will be websites, and public data spaces. But he’s thinking bigger.

“Imagine I have a hologram somewhere in Ikeja [capital of Lagos] mounted in one of the bus stops,” he says, describing a futurist scenario where anyone can access a portal of information—a digital public library. 

“This could be culture-related, economics-related, geography-related, or medicine. It could be anything, but it’s rooted within our heritage and our cultures.”

The model would be both public and private, but accessible to everyone. And critically, it would compensate contributors. 

“I want to create a platform where you could contribute that data, and on one side of the platform, you’d also be earning some sort of royalty if your data is used for anything,” he explains.

This stands in stark contrast to how AI companies typically operate. “We’re trying to change that,” Afegbua says. “I’m trying to work on a model where everybody has some sort of stake in it.

The work that pays for the dream

To fund his cultural preservation work, Afegbua runs Sleek City Media (production), Sleek City XR (virtual and augmented reality), and Sleek City.io (AI and tech). His client list includes American Express, GTBank, Access Bank, and Nando’s. He hasshot commercials, music videos, and produced content for international media companies like Babel.

His core team is small—three to five people—but expands dramatically depending on projects, sometimes reaching 50 to 100 people for major commercial productions. 

What loss looks like

Afegbua is also working on Remembery, a project using AI to reimagine heritage sites lost to natural disasters, war, or lack of maintenance. Using drawings, descriptions, and whatever archival data exists, he recreates these spaces in virtual reality so people can visit them in the metaverse.

For Afegbua, these are not just creative projects; they’re urgent interventions.

The stakes become clearer when you consider the scale of what’s being lost. Nigeria has over 500 languages, many with fewer speakers each generation. 

Cultural practices disappear when elders die without passing down knowledge. And in the age of AI, whoever controls the data controls the narrative.

“The bigger picture is to restore lost languages, to capture our actual languages, culture, heritages, so the world understands who we are, where we’re coming from, without misrepresenting it,” Afegbua says.

The question of control

As AI becomes more central to African life, the question of who owns African data becomes existential. 

Afegbua is adamant, “I don’t want to sell this data to any big tech. It has to be controlled here.”

His vision is to build and innovate locally, creating products and solutions from African data in finance, medicine, and other sectors. 

It’s a tricky balance. His work requires capital—for data storage, for human resources, for technology infrastructure. Local data storage alone represents a significant investment. Yet accepting funding from Big Tech or international organizations could compromise the project’s independence.

For now, he moves forward drop by drop, collaboration by collaboration, interview by interview with elders before they’re gone.

What comes next

Afegbua is launching a toy collection rooted in African heritage and ancestry later this year. He’s producing a TV series based on the stories he’s collected, set to air in 2026. He’s showcased his Elder Series—AI-generated fashion featuring older Africans—in Amsterdam, Lagos, and Milan, with upcoming shows in Barcelona and Casablanca, sponsored by organizations like Mozilla Festival.

But Legacy Link remains his north star. He has secured interviews with prominent scholars, including a professor who was classmates with Chinua Achebe. He’s developing templates so that cultural documentation doesn’t have to be centralized—other creators and communities can contribute to the growing database.

“I want this to be some sort of natural occurrence as a creator, as a human being, to understand that your stories have to be intentional, have to be captured, and have to be controlled by you,” he says.

For a man who started by selling T-shirts and ended up trying to make ancestors speak to the future, the journey has been anything but conventional.