The (Gen)erative (A)verage (I)ndustry
| Article | |
|---|---|
| Title | The (Gen)erative (A)verage (I)ndustry |
| Author(s) | David de Hilster |
| Published in | NLU Global |
| Date | 2026-02-15 |
| Subject area | Artificial Intelligence |
| Topics | artificial intelligence, large language models |
Read online | Read the full article |
Archived copy | Archived version |
| Retrieved | 2026-07-23 |
"The (Gen)erative (A)verage (I)ndustry" is an article by David de Hilster published on 15 February 2026 on NLU Global.
De Hilster builds his argument from a statistical curiosity: averaging photographs of many faces judged unremarkable produces a face people find beautiful. His explanation is that a lifetime of seen faces trains a recognition network, and the closer a face sits to that average the more strongly it fires. He applies the same reading to generative models — what they produce is the average of their training material, which is precisely why it is comfortable to look at and why it is not new.