The fundamental problem with any intervention that tries to eliminate certain behaviours from an LLM is that it creates incentives for the model to develop workarounds that preserve those behaviours, while evading detection. The machine simply learns to put on a false face. To be clear, these models don't ‘want’ to deceive us. They have no desires or intentions at all. They’re just doing whatever works best to accomplish their assigned tasks. The AI follows the path of least resistance through the “environment” we create for it.
Karen Swallow Prior is working her way through Paradise Lost at her Substack, and graciously asked me if I wanted to write on any one of the twelve books. I chose Book X. Thanks for the invitation, Karen!
Learned this past weekend that Half Price Books pays even less for DVDs than it does for books. I didn’t know that was possible.
John Betjeman, from “The Conversion of St. Paul”
John Ruskin, “Study of Dawn: the first Scarlet on the Clouds”
I responded to a Robin Sloan post by advocating a “distributed localism.”
Rachel Ruysch, “Posy of Flowers, with a Beetle, on a Stone Ledge” (1741)
Keita Morimoto, “Crossroad” (2025), acrylic and oil on linen