The resursive bias machine

08:03 · 22nd Jul 2026

The issue, no, a issue I have with large language models (LLM/AI) is that it isn’t the sum knowledge of human experience to start with. It’s the corpus of the people who had the power to write, so we need to examine the history of literacy as a whole. Then we need to understand Who is training (ie, restricting what an LLM is allowed to generate) the output? What are their points of views. What we have today, is a recursive bias machine. A bias(bias(bias)) machine if you will.