Revolt of the reader
Archived — this story has rotated out of today’s deck. It is kept here in full.
The gist
Readers revolt against LLM-authored writing, with 78% stopping immediately upon detection.
Authenticity matters more than polish, as 98% prefer imperfect human writing.
Background
The 'revolt of the reader' refers to a growing backlash among readers against articles and posts that are clearly authored by large language models (LLMs). The term was coined in a blog post by Bryan Cantrill, who argues that readers can easily detect LLM-generated text and are increasingly avoiding it. A survey of 668 developers found that 78% stop reading immediately when they detect an LLM, and 71% avoid the author in the future. This trend is significant because it threatens the effectiveness of using LLMs for public writing, as readers value authenticity over linguistic perfection.
How it unfolded
- recentlyBryan Cantrill publishes a blog post titled 'The revolt of the reader,' arguing that readers are exasperated by LLM-authored pieces and can easily detect them.
- recentlyCynthia Dunlop's survey of 668 developers reveals that 78% stop reading immediately when they detect an LLM, and 71% avoid the author in the future.
- recentlyPangram Labs releases Pangram 4, an AI-text detector that Cantrill praises for its low false positive rate, suggesting improved detection capabilities.
Who’s saying what
- Author
- Bryan Cantrill argues that using LLMs to write voids the social contract between writer and reader, and that readers can and do care about authenticity.
- Survey respondents
- 78% of developers surveyed say they stop reading immediately when they detect an LLM, and 71% avoid the author in the future.
- Caution
- Some may argue that LLM detection is unreliable, but Cantrill notes that Pangram 4 has a low false positive rate, making it a useful tool.
Still unverified
The survey by Cynthia Dunlop is based on self-selected respondents, which may not represent the general population.