The 100k whys of AI
Points and comments are a snapshot, not live.
LLM-generated Amazon books become indistinguishable through quasi-deterministic outputs like identical titles and cover designs.
The author argues that LLM text is distinguishable from human writing not by individual mannerisms but by quasi-deterministic homogeneity across many outputs. A collage of 220 Amazon book covers from a search for "100000 whys" reveals near-identical titles, recurring cover art (T-Rex on the left, red-and-white rocket, golden retriever), and repetitive author names (e.g., Ethan Bright, Nolan Bright). The author contends this uniformity arises because many users feed similar prompts to a small number of models, producing functionally identical output roughly 80% of the time.
What commenters are saying
Commenters broadly agree that LLM outputs converge due to training on similar data and lack of diverse life experience or mood. Top comment notes LLMs represent only 3–5 models versus 1,000 unique humans. Dissenters argue that extensive system prompts, context, and controlled randomness can yield variety, with one commenter outlining a multi-step workflow for generating distinct books. Others counter that prompt engineering is largely pseudoscience and that LLMs inherently produce average, unsurprising copy. A commenter reference the paper "TinyStories" as a related research project.