Crowdsourcing Artificial Intelligence in Africa:  Analysis of a Data Science Contest

30 Pages Posted: 5 Apr 2022

See all articles by Wim Naudé

Wim Naudé

RWTH Aachen University; IZA Institute of Labor Economics; ASC University of Leiden; University of Johannesburg

Amy Bray

Zindi

Celina Lee

Zindi

Abstract

We study the crowdsourcing of Artificial Intelligence (AI) through a data science contest on the largest intermediated digital contest platform in Africa. In total, 614 contestants submitted 15,832 entries. Analysing these entries we answer several questions: who take part in these contests and why? Who is most likely to win? We find firstly that the profile of who participates in the data science competition is of relatively young people, with only intermediate to beginner level skills in AI, and mostly not formally qualified in data science. Secondly, we find that participation is mainly driven by the intellectual challenge, the opportunity to gain experience in coding and the opportunity to showcase expertise. Thirdly, we find that experience and skills matter most for performing well. Participants based in Africa seem, on average, to obtain lower results. Our empirical results suggest that data science contests may help reduce labor market and educational obstacles to the adoption of AI. Recommendations for further research are made.

Keywords: Crowdsourcing, innovation, data science, artificial intelligence, Africa

Suggested Citation

Naudé, Wim and Bray, Amy and Lee, Celina, Crowdsourcing Artificial Intelligence in Africa:  Analysis of a Data Science Contest. Available at SSRN: https://ssrn.com/abstract=4076351 or http://dx.doi.org/10.2139/ssrn.4076351

Wim Naudé (Contact Author)

RWTH Aachen University ( email )

Templergraben 55
52056 Aachen, 52056
Germany

IZA Institute of Labor Economics ( email )

P.O. Box 7240
Bonn, D-53072
Germany

ASC University of Leiden ( email )

Netherlands

University of Johannesburg ( email )

South Africa

Amy Bray

Zindi

Celina Lee

Zindi

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
49
Abstract Views
267
PlumX Metrics