African Voices Focus AI Risk Debate on Misuse, Dependence and Governance

As artificial intelligence systems become more capable of reasoning and completing tasks with less human input, eight people from across five groups have offered differing views on whether the technology could eventually threaten humanity.

The participants, drawn from everyday users, educators, innovators, policy specialists and investors, generally rejected the idea that human extinction by AI is inevitable. Their concerns focused instead on dependence, misinformation, cybercrime, surveillance, labour displacement, bias, poor governance and systems that may act without following human values.

Concerns about everyday use and education

Sarah Pra, a restaurant owner in Kumasi, Ghana, said she uses Meta AI and ChatGPT for cooking guidance, restaurant flyers and personal questions. However, she said she does not fully trust their answers after encountering incorrect information.

“Which AI is going to kill us exactly? Is it ChatGPT or Meta AI?” Pra asked. She also pointed to the creation of fake nude images of women and the spread of false information as concerns, while attributing the danger primarily to how people use the technology.

Elijah Akuma, a master’s student, writer, researcher and teacher at the University of Nigeria, Nsukka, said he uses AI to test ideas, critique drafts and suggest improvements. He stressed that the technology does not replace his own thinking.

“AI doesn’t think for me,” Akuma said. Although he regarded the idea of AI killing everyone as unrealistic, he warned that students could become dependent on it without engaging their own ingenuity.

Aniedi Akpan, Head of Careers and Team Lead for Relationship Management at 16 Plus Schools, said AI could support personalised learning by offering explanations, practice and feedback. But he warned that students may complete assignments without gaining the underlying knowledge.

“The important question is becoming less ‘Can you find the answer?’ and more ‘Do you understand the answer, and can you determine whether it is correct?’” Akpan said. He also cautioned that AI can be confidently wrong, reproduce bias and provide convincing information without genuine understanding.

Human control remains central

ThankGod Obiechina, a business growth strategist at Wehostname.com, said AI could threaten humanity without necessarily killing everyone. He expressed concern that excessive dependence could weaken reasoning, logic and human interaction.

Abiodun Adetona, an AI developer and founder of Decide AI, identified job displacement and the ability of bad actors to use AI against fragile systems as serious risks. He argued that development should continue, but with tighter controls over access and the actions AI systems can take.

“We shouldn’t stop building because there are risks. But moving fast doesn’t mean being careless,” Adetona said. He added: “The more consequential the action, the stronger the controls around it should be.”

Kelvin Roman, founder of Daraja AI, said the greatest risks depend on how humans design, deploy and govern AI. He said Daraja AI uses safety checks, testing protocols and transparent oversight, and does not allow its Babel AI system to generate harmful content, violate privacy or support illegal activities.

“The real risk comes from how humans design, deploy, and govern these systems,” Roman said. He described bias, misinformation and critical errors at scale as more immediate dangers than scenarios in which AI independently becomes hostile to humanity.

Economic potential and alignment risks

Pius Bankong, an investment associate at Oui Capital, described AI as a force multiplier that could generate economic growth, expand global GDP and enable activities that were previously difficult or unavailable. He said corporate governance, public policy, safeguards and long-term thinking would be needed to manage the risks.

Kayode Adeniyi, an AI policy expert and researcher pursuing a PhD at Imperial College London, highlighted the challenge of AI alignment: ensuring that advanced systems follow human values and intentions. He also pointed to rapidly generated misinformation and convincing fake images and videos.

Adeniyi said an advanced system that is not aligned with human values could produce harmful outcomes even while attempting to solve a human problem. He argued that African governments are paying more attention to adoption and infrastructure than to safety and evaluation, and called for greater expertise in AI safety, research, engineering and evaluation.

The report also cited a 2026 study by Patience Ebekeyi Bakre and Fa’iza Tukur-Bello, published by Springer in Discover Education. The study surveyed 535 students across eight Nigerian tertiary institutions about uses of ChatGPT, Grammarly and Gemini, including coding, essay writing, finding materials and summarising information.

Across the interviews, the prevailing position was that AI development should not be abandoned. Instead, the participants supported stronger safeguards, responsible use, oversight and governance as the technology continues to advance.