Small Language Models (SLMs) offer Nigeria a viable route to enhance AI innovation by providing accessible, cost-effective solutions for local needs. Experts emphasize their potential to bridge the digital divide, support mobile economies, and facilitate integration into workflows, despite the limitations faced compared to Large Language Models. The development of SLMs can significantly empower Nigeria’s AI sector and foster local technological advancement.
Small Language Models (SLMs) are emerging as a viable pathway for Nigeria and Africa to enhance Artificial Intelligence (AI) innovation. Experts assert that, since the debut of ChatGPT in November 2022, the development of Large Language Models (LLMs) has showcased the vast potential of AI, influencing subsequent models such as Google’s Gemini and Microsoft’s Co-pilot. These developments have led to various applications in generative text, speech, images, and videos.
However, LLMs require significant computational infrastructure and large datasets, creating accessibility challenges in regions like Nigeria. Olubayo Adekanmbi, founder of Data Science Nigeria, alongside AI expert Ife Adebara, highlighted in a white paper the detrimental effects of LLMs’ inaccessibility. As noted in a World Economic Forum article, while LLMs like GPT-4 have over 175 billion parameters, SLMs operate with tens of millions to under 30 billion parameters, making them more approachable.
The energy requirements of LLMs are impractical for areas with limited infrastructure, such as Nigeria. The nation’s draft AI strategy identified the urgency of building affordable, localized infrastructure to support AI growth. Olivia Shone, senior director at Microsoft, pointed out that SLMs focus on specific tasks that consume fewer resources, thus being more accessible and economically viable.
According to Adekanmbi and Adebara, SLMs mitigate barriers for governments and small businesses in integrating generative AI into their operations. They stated that SLMs require less computational power, provide lower latency, and allow efficient customization. Furthermore, SLMs can significantly reduce the digital divide, facilitating AI access to underserved populations.
Both experts emphasized the importance of SLMs in mobile-driven economies like Nigeria, noting their offline usability. Libing Wang from UNESCO and Tianchong Wang from Swinburne University mentioned that SLMs represent a strategic solution for overcoming challenges in digital infrastructure within Global South countries by utilizing local expertise.
Despite these benefits, the World Economic Forum cautioned that SLMs have limitations, including reduced accuracy in complex tasks and a narrow operational scope. Nonetheless, their potential for shaping a more accessible AI landscape is considerable, particularly in contexts that prioritize efficiency and affordability.
In conclusion, Small Language Models present a significant opportunity for Nigeria and Africa to foster AI innovation by overcoming the limitations posed by Large Language Models. The potential benefits of SLMs include increased accessibility, cost-effectiveness, and adaptability for local needs, particularly in mobile-driven economies. While challenges remain regarding complex language tasks and performance, the strategic development of SLMs can substantially contribute to the digital transformation across various sectors in Nigeria.
Original Source: businessday.ng
