On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?
Connects language-model development with environmental costs, training-data bias and the risks of interpreting fluent text as meaningful communication.
What it contributes
Brings several strands of risk literature together to challenge the assumption that scaling language models is an unqualified improvement.
Read with care
A critical synthesis and argument, not an experiment establishing the behaviour of every current model. Its claims should be read alongside subsequent empirical work.
Why it belongs here
A starting point for examining whose language is represented, who bears costs and what users infer from an apparently understanding reply.
Published at the 2021 ACM Conference on Fairness, Accountability, and Transparency. Author names follow the published paper.
Source reading depthAuthor's publication page, publisher metadata and the University of Washington's account of the paper checked. This is a source note, not a full critical review.
6,706 citations · OpenAlex ↗
Snapshot: 1 October 2026. Citation counts indicate attention, not reliability; counts differ across databases.