myna23/llm-trust-africa

# 🤖 LLM-Trust-Africa Dataset > **"Do You Believe the Machine? Human Trust and Perception of LLM-Generated Content Across Twelve African Countries"** > *Supplementary annotated dataset for empirical research on human trust in LLM-generated content* --- ## 📋 Overview This repository contains the **publicly available annotated dataset** of **1,236 human–LLM text pair evaluations** collected in the first large-scale empirical study of how people across **12 African countries** perceive and trust content generated by Large Language Models (LLMs). The dataset accompanies the paper: > **Do You Believe the Machine? Human Trust and Perception of LLM-Generated Content Across Twelve African Countries** > Amina Salifu > Pre-registration: OSF `osf.io` --- ## 🌍 Why Africa? The vast majority of prior trust-in-AI research has drawn samples from **Western, Educated, Industrialised, Rich, and Democratic (WEIRD)** countries — primarily the United States, United Kingdom, and Western Europe. This dataset is, to our knowledge, the **first of its kind to systematically measure LLM trust perception across diverse African populations**, including participants from: | Country | Region | n | |---|---|---| | Nigeria | West Africa | 88 | | South Africa | Southern Africa | 73 | | Kenya | East Africa | 60 | | Ethiopia | East Africa | 40 | | Ghana | West Africa | 37 | | Tanzania | East Africa | 30 | | Uganda | East Africa | 21 | | Zambia | Southern Africa | 17 | | Rwanda | East Africa | 12 | | Senegal | West Africa | 12 | | Zimbabwe | Southern Africa | 12 | | Cameroon | Central Africa | 10 | Topics in the stimulus corpus are **African-contextualised**, covering the African Union Climate Summit, AfCFTA trade, Sahel food security, the H3Africa genomics initiative, East African data protection law, and more. --- ## 📁 Repository Structure ``` llm-trust-africa/ │ ├── participants.csv ← 412 rows, one per participant, all demographics ├── trials.csv ← 1,236 rows, one …