Blog by Anna Colom, Marta Poblet, and Stefaan Verhulst: “Independent and public interest media have for long been considered a key public good for societies, a pillar of democracy and accountable governance. Well funded public media systems have been found to consistently correlate with ‘healthy democracies’. However, the struggle for public interest media as a resilient sustainable public good is a long-standing one. Decades-long market fragmentation, the digital transformations turning the public sphere into multimodal networked spheres, and more recently, the entering into the mainstream of generative AI (GenAI) systems, and General Purpose AI more broadly, have kept public interest and independent media in a constant state of survival and reinvention.
The current trends in AI development -particularly large GenAI models- risk creating a vicious cycle for journalism and public interest media. These models are trained on high-quality journalistic content but the methods for capturing the data have been associated with exploitative and extractive practices of web scraping, routinely disregarding property rights, licensing, quality, (mis)representation and bias as GenAI models often use these data without providing provenance, context or integrity. Once deployed, they often draw audiences away from the very outlets that produced this content by providing AI-generated summaries and news-like outputs directly to users. As a result, media organisations lose readership, along with the advertising and subscription revenue that sustains independent reporting. With fewer resources, their ability to produce quality journalism declines, which in turn reduces the availability of trustworthy content. This cycle threatens not only the financial viability of public-interest media but also the integrity of the AI systems that depend on their work, and ultimately, the integrity of the available knowledge to the public…(More)”.
