Press
Amsterdam, 8 Oct 2025. In this opinion piece published on AG Connect, Rewire's Simon Koolstra warns of "model collapse" — a process in which AI models trained on AI-generated content produce increasingly generic, error-prone, and biased output. As more online content is created by AI and fed back into training data, the feedback loop erodes quality and diversity in ways that are already visible in stock imagery, SEO copy, and product reviews. The antidote, Koolstra argues, is deliberate human input: curating datasets, annotating content, and critically reviewing AI output rather than accepting it unchecked. Developers, businesses, and policymakers all have a role to play in maintaining the quality of our digital information environment — before the consequences become impossible to ignore.
