New essay!

The thesis is that Large Language Models (LLMs) like ChatGPT, Gemini, Claude, and even Grok are a kind of anti-social media, pushing back against the various properties and pathologies of social media platforms.

Whereas social media has been democratising, epistemically diverging, engagement-optimised, and performative, LLMs are “technocratising”, epistemically converging, accuracy-optimised, and polite.

From experience, I know that lots of people strongly disagree with this analysis, so I’m looking forward to the objections!