Anti-Greenwashing
Published in Private collaboration between Open Universiteit and APG, 2023
A private collaboration between Open Universiteit and APG (2023–2025) on AI-based detection of corporate greenwashing. Grounded in a systematic literature review (PRISMA methodology) covering how greenwashing is defined, what datasets and AI techniques existing solutions use (transformer-based classifiers such as BERT, RoBERTa, and ClimateBERT), how they’re evaluated, and what technical, governance, and ethical recommendations follow.
With Deniz Iren, Clara Maathuis, Gerard Mertens, Karen Maas, and Petru Curseu at Open Universiteit.
Technology: Reproduced Bingler et al.’s Cheap Talk Index (2024), which scores specificity, commitment, and action in corporate climate disclosures via ClimateBERT classifiers, then built a Python pipeline (Hugging Face transformers, spaCy, Azure Document Intelligence for PDF paragraph extraction) and a web interface to operationalize it, tested on real annual reports (ASML, Audi, Heineken, H&M).
My role: Reproduced the Cheap Talk Index methodology and built the pipeline and web interface end to end.