Position: the Stochastic Parrot in the Coal Mine. Model Collapse is a Threat to Low-Resource Communities

TL;DR

This paper discusses how model collapse in generative models harms low-resource communities by reducing training efficiency and reinforcing biases, emphasizing environmental and cultural impacts.

Contribution

It highlights the threat of model collapse to democratizing AI, especially affecting marginalized groups, and proposes initial mitigation strategies.

Findings

Model collapse reduces training efficiency.

It skews data distribution away from rare patterns.

It disproportionately impacts low-resource communities.

Abstract

Model collapse, the degradation in performance that arises when generative models are trained on the outputs of prior models, is an increasing concern as artificially generated content proliferates. Related critiques of large language models have highlighted their tendency to reproduce frequent patterns in training data, their reliance on vast datasets, and their substantial environmental cost. Together, these factors contribute to data degradation, the reinforcement of cultural biases, and inefficient resource use. In this position paper we aim to combine these views and argue that model collapse threatens current efforts to democratize AI. By reducing training efficiency and skewing data distributions away from the tails of their support, model collapse disproportionately impacts low-resource and marginalized communities. We examine both the environmental and cultural implications of this phenomenon, situate our position within recent position papers on model collapse, and conclude with a call to action. Finally, we outline initial directions for mitigating these effects.