An Artificial Intelligence (AI) Educator, Ms Zainab Saidu , says most AI tools still struggle with Nigerian languages and contexts, warning that reliance on Western-trained systems risks bias and misinformation.
Saidu, who is also the Founder of Creativity Enthusiast Network, said this in an interview with the News Agency of Nigeria (NAN) while assessing the performance of current AI systems in Nigeria.
NAN reports that AI is the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning problem solving, perception and decision-making.
The expert, therefore, explained that many AI tools often misinterpret Hausa, Yoruba, Igbo and Nigerian Pidgin, leading to errors in tone, meaning and cultural interpretation.
She said “AI systems get confused by proverbs, dialect variations and everyday expressions that are normal to us. Speech recognition also struggles with accents and background noise, although there are gradual improvements due to local projects collecting Nigerian speech data.”
Saidu warned that heavy reliance on AI trained mainly on Western datasets poses significant risks, including cultural misrepresentation, bias and misinformation, particularly in sensitive sectors such as healthcare, governance and journalism.
According to her, symptoms described in Pidgin or Hausa can be wrongly interpreted by AI tools because they are trained largely on Western medical data, potentially leading to harmful outcomes.
She noted that for AI to truly serve African journalism, stakeholders must prioritise the development of local datasets, including newsroom interviews, reports and real-life conversations that reflect code-switching and cultural nuances.
“This will require collaboration among media organisations, technologists, linguists, universities and government, as well as sustained funding and ethical, transparent processes,” she said.
Saidu acknowledged that Nigeria still depends heavily on Western technology firms but said the narrative was gradually changing, with government agencies such as the National Information Technology Development Agency (NITDA) and startups like Awarri and CDIAL AI investing in local language models.
She stressed the need for clear ethical guidelines, transparency in AI use, journalists’ training, fact-checking systems and strong data protection policies to prevent misuse in the media.
“Safeguards should not stifle innovation but ensure AI strengthens public trust in journalism,” she added.(NAN)