By Milcah Tanimu
Dr. Oluseyi Akindeinde, founder of Hyperscale and the driving force behind Neural AI, has highlighted the prohibitive costs associated with training data as a significant obstacle to the advancement of Artificial Intelligence (AI) in Nigeria.
Speaking at the Nigerian Information Technology Reporters Association (NITRA) ICT Growth Conference 4.0 in Lagos, Akindeinde pointed out that many Nigerian companies with AI ambitions are struggling due to the high expenses involved in training AI models.
“Training AI requires extensive servers, which are prohibitively expensive. For example, NVIDIA, a major player in AI, sells servers for around $500,000 each. With multiple racks needed for effective training, the total cost quickly escalates to millions of dollars,” Akindeinde explained.
AI Education and Patenting Challenges
Akindeinde also noted a significant gap in AI education within Nigeria, which is impeding the progress of AI technology in the country. He emphasized the need for broader understanding and education about AI, especially among legal professionals involved in patenting innovations.
“It is crucial to educate lawyers, judges, and other non-technical individuals on AI and blockchain technologies to ensure informed decisions in the patenting process. This knowledge gap is not only a technical issue but also an educational one,” he said.
Policy and Collaboration for AI Advancement
Mr. Adesola Akinsanya, President of the Nigeria Internet Registration Association (NiRA), also addressed the conference, stressing the transformative potential of AI. However, he noted that realizing these benefits requires careful planning, sound policies, and collaborative efforts.
“AI is reshaping industries worldwide, and Nigeria must not miss out. Addressing the challenges and seizing opportunities will require collective action, partnerships, and shared knowledge,” Akinsanya stated. He urged stakeholders, including industry leaders, policymakers, and academia, to collaborate with NITRA and NiRA to foster a robust digital ecosystem.
Global and Local Context
A recent Stanford University report highlighted the escalating costs of training AI models, which are limiting the participation of non-industry actors in the tech revolution. The Artificial Intelligence Index Report 2024 noted that training expenses can reach millions of dollars, citing OpenAI’s $100 million expenditure on GPT-4 as an example.
In response, the Nigerian government has launched its Large Language Model (LLM) project, aimed at positioning Nigeria as a leader in AI within Africa. Dr. Bosun Tijani, Minister of Communications, Innovation, and Digital Economy, revealed that the LLM will include training in five low-resource languages and accented English to enhance the representation of diverse languages in AI datasets.
