The risks of AI slop and AI model collapse, and why it is essential to adequately feed the next Generative AI models and to remunerate creators through a dual right system
This article addresses the twin and related challenges posed by generative AI: the rise of low-quality AI-generated content (‘AI slop’) and the risk of model collapse due to overreliance on synthetic data, often less diverse and rich than human-created content. Both trends threaten the sustainabi...
Gespeichert in:
| 1. Verfasser: | |
|---|---|
| Format: | Article ou chapitre numérique |
| Sprache: | Français |
| Veröffentlicht: |
2025
|
| Online Zugang: | Accès Université d'Orléans et IFPM Accès Université d'Orléans et IFPM |
| Zusammenfassung: | This article addresses the twin and related challenges posed by generative AI: the rise of low-quality AI-generated content (‘AI slop’) and the risk of model collapse due to overreliance on synthetic data, often less diverse and rich than human-created content. Both trends threaten the sustainability of generative AI development and highlight the urgent need for high-quality human-created content. The article argues for a recalibration of copyright law to incentivise human creativity through a dual rights system : (1) facilitating the exercise of exclusive rights via a trusted copyright infrastructure relying on the standardisation of copyright metadata and (2) introducing an unwaivable remuneration right for creators and performers. Such a model is essential not only to preserve fair compensation for the creative community but also to ensure the availability of the diverse, high-value data that AI systems critically depend on. The article outlines legal and policy pathways – including statutory licensing, output-based levies, and extended collective licensing – to adapt copyright to the evolving AI landscape.
|
|---|