The logical destination of LLMs

This article discusses Graze's announcement of co/core, an AI cooperative aiming to 'billionaire proof inference' by allowing users to share spare compute from devices like old MacBooks to run LLM tasks. The author argues that the logical destination of LLM technology is small, niche networks and local models used by enthusiasts, not mass adoption, due to poor economics and ROI. While appreciating the initiative's move away from tech oligarchy, the author expresses personal disinterest due to environmental, psychological, and societal harms. The article also includes a personal introduction about the author's work in web development and CSS education.