The Founder of Hyperscale, the company behind Neural AI, Dr. Oluseyi Akindeinde, has identified the high cost of training data as a major factor that may slow down the development of Artificial Intelligence (AI) in Nigeria.
Akindeinde stated this during the Nigerian Information Technology Reporters Association (NITRA) ICT Growth Conference 4.0 held in Lagos.
According to him, many Nigerian companies that have AI innovations are unable to train their own data because it is quite expensive.
“To train AI data, you need huge servers and these servers are costly. NVIDIA is the biggest company in AI today because they sell these servers that are used to train AI data and one of them is about $500,000 and you need about two to three racks of about 10-10 of them. So, we’re talking about millions of dollars,” he said.
AI education in Nigeria
While noting that the understanding of AI technology is still very low, Akindeinde said the country would need a lot of AI education across sectors to make headway with the technology.
He observed it is currently difficult to get any AI patented in Nigeria because those who are in charge do not understand the technology. According to him, Nigerian lawyers in particular, and other non-technical people that are involved in the process of patenting innovations need to be educated on the technology.
“We need to educate lawyers, judges, and others about blockchain, generative AI, and other technologies, so they can make informed decisions.
‘’We need to address this education gap to move forward and leverage AI effectively. It’s a technical issue, but also a matter of education and awareness,” he said.
AI policy
Also speaking at the event, the President of Nigeria Internet Registration Association (NiRA), Mr Adesola Akinsanya, said that AI is rapidly transforming industries and economies across the globe, and Nigeria should not be an exception.
He, however, noted realizing the benefits requires careful consideration of the prospects, the formulation of sound policies, and addressing the accompanying challenges.
“No single entity can address the complex challenges we face alone. It is through partnerships, shared knowledge and collective action that we can drive the kind of growth that is inclusive, sustainable, and transformative,’’ he said.
- Akinsanya called on all stakeholders, industry leaders, policymakers, academia and civil society to join hands with NITRA, even as NiRA was committed to the cause of building a robust digital ecosystem.
- According to him, there is a need to work together to create policies that support innovation and develop infrastructure that meets the demands of tomorrow.
- He added that it was also necessary to cultivate a workforce that was skilled and ready to lead in a digital world.
What you should know
A recent report by Stanford University had also revealed that the high cost of training AI models is limiting the involvement of non-industry actors in the technology revolution.
- According to the Artificial Intelligence Index Report 2024, while AI companies seldom reveal the expenses involved in training their models, it is widely believed that these costs run into millions of dollars and are rising.
- Citing openAI as an instance, the company’s CEO, Sam Altman, last year mentioned that the training cost for GPT-4 was over $100 million.
- According to the Minister of Communications, Innovation and Digital Economy, Dr. Bosun Tijani, the LLM will be trained in 5 low-resource languages and accented English to ensure stronger language representation in existing datasets for the development of Artificial Intelligence solutions.