Research
Structured, research-driven writing for deeper system-level analysis.
Algorithmic Colonialism: Why Global AI Safety Guardrails Fail the Nigerian Reality
Praise Oshin
The safety guardrails for large language models (LLMs) at scale globally are still basically tuned to Western sociological baselines. This research examines the systemic friction of these imported protocols with the Nigerian reality. This paper, through a localised lens, analyses Reinforcement Learning from Human Feedback (RLHF) and shows how global AI models often misclassify the localised Nigerian informal economy as a risk vector and misinterpret indigenous linguistic nuances such as Pidgin and localised syntax as policy violations. Ultimately, this research makes the case that true algorithmic alignment must come from indigenous red-teaming to ensure global models adapt to African realities rather than enforce digital exclusion.
The Architecture of the "Urgent 2k": Nigeria's Invisible Micro-Lending Network
Praise Oshin
This research investigates the structural mechanics of the "urgent 2k"—a ubiquitous, informal micro-lending network prevalent across Nigerian youth demographics. With tough collateral and credit history requirements, this demographic is generally excluded from formal credit systems, but has built a decentralised, zero-interest peer-to-peer lending ecosystem. This paper discusses the role of social capital as an alternative mechanism for credit scoring, the role of neobanks in facilitating high velocity liquidity, and the role of digital reputational threat as the main tool for debt recovery. The deconstruction of this cultural phenomenon shows systemic failures of formal models of financial inclusion, and at the same time resilience of indigenous financial engineering.