Title: Algorithms smarter than experts? AI methods applied to assessment of environmental risk of World Bank projects

Authors: Laurent Boinot; Jaime Diaz; Steffen Feuerpeil; David Manoury; Gayatri Rao

Addresses: Harvard Kennedy School, 79 John F. Kennedy St, Cambridge, MA 02138, USA ' Harvard Kennedy School, 79 John F. Kennedy St, Cambridge, MA 02138, USA ' Harvard Kennedy School, 79 John F. Kennedy St, Cambridge, MA 02138, USA ' Harvard Kennedy School, 79 John F. Kennedy St, Cambridge, MA 02138, USA ' Harvard Kennedy School, 79 John F. Kennedy St, Cambridge, MA 02138, USA

Abstract: Environmental and social risk assessment of investment projects can be enhanced using text analytics. For each project, the World Bank assesses environmental risks and prepares environmental and societal safeguards for mitigation before board approval. In this paper, we used data available at the time of board approval regarding the project objectives and project descriptions to predict the risk of failure. We demonstrate that a simple model of text analytics, which goes beyond regular assessment reports, unearths hidden patterns to improve the evaluation of environmental and social hazards. We recommend that the World Bank improves its non-financial risk assessment by adopting artificial intelligence techniques. We also propose that the World Bank integrates text analytics in its open data and artificial intelligence initiatives.

Keywords: environmental impact; international development; machine learning; non-financial risk assessment; big data for development; decision trees; natural language processing; text analytics; artificial intelligence; multilateral organisations; development banks.

DOI: 10.1504/IJSD.2023.133605

International Journal of Sustainable Development, 2023 Vol.26 No.2, pp.71 - 85

Received: 17 Jul 2021
Accepted: 29 Nov 2022
Published online: 25 Sep 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article