Data for the paper: From Potential to Practice: Modelling AI Readiness and Scholarly Integration in African Research Systems

Abstract

This repository contains the cleaned, harmonized analysis datasets supporting the empirical study of artificial intelligence (AI) readiness and realized scholarly AI integration across African research systems. The data integrates large-scale bibliometric metadata retrieved from the OpenAlex database (covering 43,578 cleaned, peer-reviewed AI-related publications affiliated with African institutions) with national policy metrics from the Oxford Insights Government AI Readiness Index (AIRI). The datasets capture longitudinal trends in publication intensity, time-normalized citation impact, topic diversity (Simpson Index), and intra-African collaboration networks. Methodological Notes: Bibliometric Source: OpenAlex API (Extraction date: November 2025). The corpus was rigorously filtered to include only peer-reviewed or editorially vetted outputs (journal articles, conference proceedings, and book chapters) from 2000 to 2025. Policy Source: Oxford Insights Government AI Readiness Index (AIRI) scores, strictly harmonized to a 100-point scale for the 2019–2024 econometric window. Counting Method: Country-level aggregations utilize a full-counting approach, whereby multi-country collaborative works are credited to all participating African research systems. File Descriptions: panel_with_readiness.csv: The primary country–year panel dataset (2019–2024) used for the Two-Way Fixed Effects (TWFE) panel regressions. It contains national AI readiness scores alongside the composite Scholarly AI Integration Score and its underlying pillars (log publication intensity, relative citation impact, and topic diversity) for 52 African nations. country_centrality.csv: The structural network metrics used for the intra-African collaboration analysis, containing eigenvector and betweenness centrality scores for each national research system. institution_flagship_highered_top20_2000_2024.csv: Aggregated institutional output and citation metrics detailing the extreme concentration of AI scholarship among Africa’s leading higher-education "flagships." ai_education_top15.csv: The specific thematic subset used to analyze the highly concentrated AI-in-education research domain. Usage: Researchers are encouraged to use these datasets to replicate the core fixed-effects models, recreate the median-based readiness–integration typology, or conduct comparative scientometric analyses.